Exp Physiol 91.

6 pp 967–976 967

Experimental Physiology

Effect of frequency and pulse duration on human muscle

fatigue during repetitive electrical stimulation
Trisha Kesar and Stuart Binder-Macleod
Department of Physical Therapy, 301 McKinly Laboratory, University of Delaware, Newark, DE 19716, USA

Different combinations of stimulation frequency and intensity can generate a targeted force
during functional electrical stimulation (FES). This study compared isometric performance and
muscle fatigue during repetitive stimulation with three different combinations of frequency and
pulse duration that produced the same initial peak forces: protocol 1 used long pulse duration
(fixed at 600 μs) and 11.5 ± 1.2 Hz (low frequency); protocol 2 used 30 Hz (medium frequency)
and medium pulse duration (150 ± 21 μs); and protocol 3 used 60 Hz (high frequency) and
short pulse duration (131 ± 24 μs). Twenty and 60 Hz pre- and postfatigue testing trains were
delivered at the pulse duration used by the fatiguing trains and at 600 μs pulse duration. The
percentage decline in peak force between the first and last fatiguing train of each protocol was
the measure of muscle performance. The declines in peak force of the 60 Hz testing trains were
used to measure muscle fatigue. The 20 Hz:60 Hz peak force ratio was used as a measure of low-
frequency fatigue. The results showed that protocol 1 produced the least decline in peak force
in response to the fatiguing trains, as well as the least muscle fatigue and low-frequency fatigue
when the pulse duration was maintained at the level used by the fatiguing trains. Interestingly,
protocol 2 produced the least muscle fatigue, and there were no differences in the levels of low-
frequency fatigue across protocols when a comparable motor unit population was tested using
600 μs pulse duration. The results suggest that if the frequency and intensity are kept constant
during FES, using the lowest frequency and longest pulse duration may maximize performance.
(Received 13 March 2006; accepted after revision 22 June 2006; first published online 14 August 2006)
Corresponding author S. Binder-Macleod: Department of Physical Therapy, 301 McKinly Laboratory, University of
Delaware, Newark, DE 19716, USA. Email: sbinder@udel.edu

Functional electrical stimulation (FES) uses electrical
stimulation to generate functional movements in
individuals with upper motor neurone paresis (Liberson
et al. 1961; Kralj et al. 1988). Functional electrical
stimulation can help individuals with paralysis caused by
upper motor neurone dysfunctions, such as spinal cord
injury, cerebral palsy and stroke, to regain the ability
to stand (Triolo et al. 1996), walk (Agarwal et al. 2003;
Johnston et al. 2005) and grasp objects (Mangold et al.
2005). Functional electrical stimulation has proven to
be effective for lower and upper extremity rehabilitation
in individuals with hemiplegia following stroke (Ring &
Rosenthal, 2005), and for providing exercise alternatives
for tetraplegic individuals through FES-induced rowing
(Davoodi et al. 2002; Wheeler et al. 2002) and cycling
(Gfohler & Lugner, 2000; Hunt et al. 2004). For effective
task performance during FES, it is necessary to maintain
the level of muscle force required for generation of the
FES-elicited movement.
During FES, however, skeletal muscles fatigue more
rapidly during repetitive stimulation than during
volitional contractions (Marsolais & Edwards, 1988;
Riener, 1999). Rapid fatigue during FES is thought to
result from the differences in motor unit recruitment
order, higher activation frequencies and imprecise control
of muscle force during FES compared to volitional
contractions (Peckham & Knutson, 2005). The problem
of muscle fatigue is further compounded by the fact that
paralysed muscles show greater fatigability than healthy
muscle (Gerrits et al. 1999, 2003). Muscle fatigue is an
important factor limiting the clinical use of FES (Riener,
1999). During repetitive electrical stimulation, stimulation
frequency and intensity are two primary parameters
that can be modulated to control skeletal muscle force.
Although numerous combinations of frequency and
intensity can be used to generate the required muscle force
during FES, most clinical FES systems use the minimum
frequency that can generate a fused tetanic contraction


C 2006 The Authors. Journal compilation 
C 2006 The Physiological Society DOI: 10.1113/expphysiol.2006.033886
968 T. Kesar and S. Binder-Macleod
in the muscle being stimulated and vary the intensity
to produce the desired force (Taylor et al. 1999; Weber
et al. 2005). The stimulation intensity is further increased
as the muscle fatigues (Taylor et al. 1999; Weber et al.
2005). The rationale for using low frequencies in FES
is based on the premise that higher frequencies cause
greater fatigue (Bigland-Ritchie et al. 1979; Garland et al.
1988). Although several previous studies have shown that
fatigue is a function of the stimulation frequency or the
number of pulses (Bigland-Ritchie et al. 1979; Marsden
et al. 1983; Garland et al. 1988; Binder-Macleod et al.
1998), these studies have often ignored the effect of
either the differences in initial peak force or stimulation
intensities on muscle fatigue (Binder-Macleod et al. 1995;
Russ et al. 2002b). Russ and colleagues (Russ et al.
2002c) recently showed that increasing the frequency or
number of pulses did not affect the amount of muscle
fatigue produced during repetitive isometric contractions
if the initial force produced by the stimulation trains
was controlled. Another study showed that intermittent
high-frequency stimulation produced less fatigue than
low-frequency repetitive stimulation in able-bodied and
spinal cord injured subjects (Matsunaga et al. 1999). Thus,
previous literature does not provide conclusive evidence
about the isolated effect of stimulation frequency on
muscle fatigue.
Only a few studies have investigated the relationship
between stimulation intensity and muscle fatigue. Binder-
Macleod et al. (1995) tested the rate and amount
of fatigue during repetitive stimulation of the human
quadriceps muscle with trains at stimulation amplitudes
that produced 20, 50 and 80% maximum voluntary
isometric contraction (MVIC) forces, and found less
decline in force of the fatiguing trains during repetitive
stimulation at higher compared to lower stimulation
amplitudes. In contrast, Godfrey et al. (2002) recently
showed greater declines in force due to fatigue at
high (supramaximal) compared to low (submaximal)
stimulation intensities. Thus, we also do not know which
stimulation intensity levels can help to minimize fatigue
when used for repetitive stimulation during FES.
Since the stimulation frequency (Bigland-Ritchie et al.
1979; Marsden et al. 1983; Garland et al. 1988; Binder-
Macleod et al. 1998), intensity (Binder-Macleod et al.
1995; Godfrey et al. 2002) and force generated in response
to electrical stimulation (Russ et al. 2002a,b) can affect
the amount of muscle fatigue produced during repetitive
stimulation, it is difficult to isolate the effects of stimulation
frequency versus intensity on muscle fatigue while
controlling for the force generated in response to electrical
stimulation. For FES applications, however, because the
targeted force is determined by the task requirements, it
may not be important to isolate the effects of frequency
versus intensity on muscle fatigue and performance,
but to determine the combination of frequency and
Exp Physiol 91.6 pp 967–976
intensity that can generate the targeted muscle force
while simultaneously minimizing muscle fatigue. It has
been hypothesized, but not systematically tested, that the
combination of the lowest frequency and highest intensity
may minimize fatigue when used for repetitive stimulation
(Binder-Macleod & Snyder-Mackler, 1993). No previous
study has attempted systematically to investigate the
combination of stimulation frequency and intensity that
can minimize fatigue while producing a targeted force level
during repetitive stimulation.
The purpose of this study was to determine which
combination of stimulation intensity and frequency
produces the least decline in force during repetitive
electrical stimulation, for the same initial peak force.
Both the amplitude and the duration of the stimulus
pulses can be varied to modulate the stimulation intensity
during electrical stimulation. For this study, we used
stimulus pulse duration (PD) to vary the intensity
because it was easier to control and provided a more
consistent force response from the muscle compared
to stimulation amplitude (Grill & Mortimer, 1996).
Specifically, we compared the percentage decline in
quadriceps femoris isometric muscle force produced
during repetitive stimulation with trains consisting of
three different combinations of stimulation frequencies
and pulse durations that produced the same initial
targeted peak force, as follows: protocol 1 used trains
with 600 μs pulse duration (long pulse duration), and
the frequency was varied for each subject to generate
the targeted force (low frequency); protocol 2 used 30 Hz
trains (medium frequency), and the pulse duration was
varied for each subject to generate the targeted force
(medium pulse duration); and protocol 3 used 60 Hz trains
(high frequency) and the pulse duration was varied for
each subject to generate the targeted force (short pulse
duration). Please note that the terms ‘medium pulse
duration’ and ‘short pulse duration’ were operational
definitions for the pulse durations used to generate the
targeted force using 30 and 60 Hz trains for protocols 2
and 3, respectively. In addition, the frequency and pulse
duration used during protocol 2 were similar to the
parameters commonly used in clinical FES systems (Taylor
et al. 1999; Donaldson et al. 2000; Johnston et al. 2005).
Methods
Subjects
Twelve healthy individuals (6 males and 6 females) aged
22–30 years participated in the study. The subjects had
no history of lower extremity orthopaedic, neurological
or vascular problems. The subjects were requested to
refrain from strenuous exercise for at least 48 h before
the testing sessions. The subjects signed informed consent
forms approved by the Human Subjects Review Board

C 2006 The Authors. Journal compilation 
C 2006 The Physiological Society
Exp Physiol 91.6 pp 967–976 Muscle fatigue during repetitive stimulation 969

of the University of Delaware. The study was performed
according to the Declaration of Helsinki.
Apparatus and set-up
The subjects were seated on an electromechanical
force dynamometer (KinCom III 500-11, Chattecx Corp.,
Chattanooga, TN, USA) with the back supported, hips
flexed approximately to 75 degrees and knees flexed at
90 deg (Fig. 1). Velcro straps were used to stabilize the
subjects’ upper trunk, waist and thigh. Each subject’s
ankle was stabilized with a strap placed approximately
5 cm proximal to the lateral malleolus. The isometric force
output of the quadriceps femoris muscle was recorded via
a force transducer placed against the anterior aspect of
the lower leg, 5 cm proximal to the lateral malleolus. The
subjects could see a representation of the force recorded
by the KinCom force transducer on a display screen.
Electrical stimulation was delivered via two
self-adhesive surface electrodes (Versa-Stim,
76 mm × 127 mm, CONMED Corp., New York, NY,
USA). The proximal electrode was placed over the upper
thigh, covering the proximal portion of the rectus femoris
and vastus lateralis muscles. The distal electrode was
placed over the lower aspect of the thigh, covering the
vastus medialis and distal portion of the rectus femoris.
The same experimenter placed electrodes across sessions,
and care was taken to maintain consistency in electrode
placement. A Grass S8800 stimulator (Grass Instrument
Co., Quincy, MA, USA) with a SIU8T stimulus isolation
unit was used to deliver the electrical stimulation. A
personal computer equipped with a PCI-6024E data
acquisition board and a PCI-6602 counter-timer board
was used. A custom-made switch was connected in series
with the stimulator to control the pulse duration. A
custom-written LabVIEW program was used for data
acquisition, and to control the timing and the duration of
the pulses.
Experimental procedure
Each subject participated in four sessions, with a minimum
of 48 h separating consecutive sessions. At the start of the
first session, subjects received an overview of the testing
procedures, signed the informed consent form, and were
trained to perform the MVIC test (Fig. 2). They were
seated on the KinCom, surface electrodes were attached
to the skin of the subjects’ thigh and tested for appropriate
placement. Next, the MVIC force was recorded using the
burst superimposition technique (Snyder-Mackler et al.
1994). During the MVIC test, the subjects attempted
to produce as much knee extension force as possible.
During the maximal voluntary contraction, an electrical
stimulation train (amplitude, 130 V; frequency, 100 Hz;
pulse duration, 600 μs) was delivered to the quadriceps
femoris muscle. This stimulation train or ‘burst’ was
superimposed on the volitional contraction to ensure
that the subjects were truly generating maximal force.
If the electrical stimulation train increased the force by
≤ 10%, the subject’s MVIC was recorded. If the electrical
stimulation train increased the subject’s force output by
Session 1
MVIC testing
> 48-hours interval
Sessions 2, 3, and 4
Set stimulation amplitude to generate 50% MVIC peak force
(using 60-Hz trains at 600-μs pulse duration)

Muscle Potentiation

Determine frequency or pulse duration of

fatiguing protocol to be tested during session
Protocol #1 Protocol #2 Protocol #3
Freq: Varied PD: Varied PD: Varied
PD: 600-μs Freq: 30-Hz Freq: 60-Hz
5-minute rest
Repeat muscle potentiation

Pre-Fatigue Testing Trains

Rest time between trains = 10-second
60-Hz @ 20-Hz @ 60-Hz @ PD of 20-Hz @ PD of
600-μs 600-μs fatiguing train fatiguing train

Fatiguing Trains
176 fatiguing trains
Rest time: 700-ms
Protocol #1 Protocol #2 Protocol #3

Post-fatigue testing trains

Rest time between trains = 700-ms
60-Hz @ 20-Hz @ 60-Hz @ PD of 20-Hz @ PD of
600-μs 600-μs fatiguing train fatiguing train

Figure 1. Schematic representation of the experimental set-up Figure 2. Flow chart showing the experimental protocol for the
used for testing 4 sessions
F indicates the position of the force transducer. Please see text for details.


C 2006 The Authors. Journal compilation 
C 2006 The Physiological Society
970 T. Kesar and S. Binder-Macleod
> 10%, the MVIC test was repeated after a 10 min rest.
If the subject failed to elicit a true MVIC within three
repetitions, they were not tested on that day. For each
subject, the MVIC value measured during the first session
was used to set the stimulation amplitude for all four
sessions.
The remaining three sessions involved fatigue testing.
The order of testing of the three fatigue protocols was
randomized across subjects. Only one protocol was tested
on each day. Each testing session consisted of ‘fatiguing
trains’ and pre- and postfatigue ‘testing trains’. The
‘fatiguing trains’, consisting of stimulation trains of three
different combinations of frequency and pulse duration
repetitively delivered at a rate of one train every second,
were used to fatigue the muscle and to assess the muscle’s
performance during the fatigue test. The ‘testing trains’,
consisting of stimulation trains of 60 and 20 Hz frequency
at two different pulse durations, were delivered before
(prefatigue) and after (postfatigue) the fatiguing trains.
The testing trains measured the decline in the force-
generating ability of the muscle. All fatiguing and testing
trains were 300 ms long. The duty cycles used to fatigue
the muscles (300 ms long trains with 700 ms rest time)
were similar to the activation patterns recorded in the
quadriceps muscle during normal walking (Pierrynowski
& Morrison, 1985). Owing to differences in frequencies
between the trains used for testing, however, the fatiguing
trains for each of the three protocols contained different
numbers of pulses.
The following procedures were followed during the
three fatigue sessions (Fig. 2):
Set stimulation amplitude to generate 50% MVIC peak
force. During each testing session, after applying the
electrodes, the stimulation amplitude was set to produce
50% of the subject’s MVIC force using 300 ms long,
60 Hz trains with 600 μs pulse duration. The 50%
MVIC amplitude was used because it allowed a range of
frequencies and pulse durations to be used to generate the
target force of 20% MVIC.
Potentiation. Eleven trains (770 ms train duration, 14 Hz
frequency, 600 μs pulse duration) were delivered with a 5 s
rest time between trains to potentiate the muscle (Binder-
Macleod et al. 2002).
Set frequency and pulse duration to generate 20% MVIC
target peak force. Either the stimulation pulse duration
or frequency of a 300 ms long train was varied to generate
peak force equal to 20% of the subject’s MVIC. The first
train of each fatigue protocol generated ∼20% MVIC peak
force.
Exp Physiol 91.6 pp 967–976
Potentiation. Potentiation was repeated before the fatigue
test to prevent the effects of potentiation from interacting
with fatigue (Binder-Macleod et al. 2002).
Prefatigue testing trains. Sixty and 20 Hz trains were
delivered at the same pulse duration as that used for the
fatiguing trains for that session and also at 600 μs pulse
duration. The prefatigue testing trains were delivered with
a rest time of 10 s.
Fatiguing trains. After the potentiation and prefatigue
testing trains had been delivered, 176 fatiguing trains were
delivered at a rate of one train every second (train duration,
300 ms; rest time, 700 ms). Three different combinations
of frequency and pulse duration were used during the three
different fatigue protocols, as follows:
Protocol 1: Long pulse duration (600 μs) and low
frequency. The pulse duration of a 300 ms long train was
fixed at 600 μs, and the frequency was set to produce 20%
MVIC peak force.
Protocol 2: Medium frequency (30 Hz) and medium pulse
duration. The frequency of a 300 ms long train was fixed
at 30 Hz, and the pulse duration was set to produce 20%
MVIC peak force.
Protocol 3: High frequency (60 Hz) and short pulse
duration. The frequency of a 300 ms long train was fixed
at 60 Hz, and the pulse duration was set to produce 20%
MVIC peak force.
Postfatigue testing trains. At the end of the fatiguing
trains, postfatigue testing trains were delivered in the same
order as the prefatigue testing trains at a rate of one train
every second to maintain the state of muscle fatigue.
Data analyses
The decline in force generated in response to the fatiguing
trains for each protocol was used as a measure of the
‘muscle’s performance’ or the muscle’s ability to maintain
force output in response to the fatiguing trains. The
percentage declines in peak force between the first and last
fatiguing train were calculated for each fatigue protocol.
Since muscle fatigue was the primary focus of our study,
it was important to define muscle fatigue in the context
of this study. Muscle fatigue is a decline in the force-
generating ability of the muscle as a result of recent activity
(Edwards, 1981; Vollestad, 1997). We used the percentage
decline in peak force between pre- and postfatigue 60 Hz
testing trains as a measure of the decline in the force-
generating ability (Vollestad, 1997). The decline in peak

C 2006 The Authors. Journal compilation 
C 2006 The Physiological Society
Exp Physiol 91.6 pp 967–976 Muscle fatigue during repetitive stimulation 971

Table 1. The sex (M = male, F = female), age, and maximal voluntary isometric contraction (MVIC) force for individual subjects; the
frequency and pulse durations (PD) used for repetitive stimulation during the 3 fatigue protocols are also listed
Protocol 1 Protocol 2 Protocol 3
Subject Sex Age (years) MVIC (N) PD (μs) Frequency (Hz) PD (μs) Frequency (Hz) PD (μs) Frequency (Hz)
1 M 22 1012 600 11.8 148 30 138 60
2 M 26 881 600 11.7 178 30 164 60
3 M 27 1117 600 14 112 30 106 60
4 M 24 1231 600 10.8 182 30 166 60
5 M 24 1335 600 12 155 30 139 60
6 M 30 1277 600 13.1 180 30 78 60
7 F 23 449 600 10 123 30 121 60
8 F 24 798 600 10.6 148 30 137 60
9 F 23 775 600 12.4 147 30 128 60
10 F 22 969 600 10.6 141 30 129 60
11 F 27 549 600 10.4 147 30 135 60
12 F 26 1000 600 10.3 145 30 137 60
Mean — 24.8 949.4 600 11.5 150 30 131 60
S.D. — 2.4 276.2 — 1.2 22 — 24 —

force of the 60 Hz testing trains at the same pulse duration Results

as the fatiguing trains provided a measure of the amount
of muscle fatigue produced in the population of motor
units recruited by the fatiguing trains during each fatiguing
protocol. The decline in peak force of the 60 Hz testing
trains at the 600 μs pulse duration provided a measure of
the muscle fatigue produced within a comparable number
of motor units as were recruited during the protocol
that used the longest pulse duration (i.e. protocol 1). In
addition, the ratio of peak forces produced in response
to 20 versus 60 Hz testing trains (20 Hz:60 Hz peak force
ratio) was used as a measure of low-frequency fatigue
(Vollestad, 1997; Russ & Binder-Macleod, 1999). The
20 Hz:60 Hz peak force ratios were calculated at the
beginning (prefatigue) and end (postfatigue) of the fatigue
protocols both for testing trains at the pulse duration of
the fatiguing trains and at the 600 μs pulse duration.
Statistical analyses
The percentage decline in peak forces from the first to
last fatiguing trains, percentage decline in peak force
between pre- and postfatigue 60 Hz testing trains at
the pulse duration of the fatiguing trains, and the
percentage decline in 60 Hz testing trains at 600 μs
pulse duration were compared using one-way repeated
measures ANOVAs. Pairwise post hoc comparisons using
the least squared difference (LSD) were performed
only if the ANOVA showed significant differences. The
pre- and postfatigue 20 Hz:60 Hz peak force ratios for
the three fatigue protocols were compared using two-
way (protocol × fatigue) repeated measures ANOVAs.
In addition, peak forces produced in response to the
first fatiguing train of each protocol were compared
using a repeated-measures one-way ANOVA to determine
whether the fatigue protocols produced similar initial peak
forces. The significance level was set at P = 0.05.
Data were collected from 12 healthy individuals (6 males
and 6 females). All 12 subjects were successfully able to
complete the MVIC testing during the first session in one
or two attempts. Table 1 provides detailed information
about the subjects’ age, sex, MVIC forces, and the
stimulation parameters of the fatiguing trains. The
frequency used for repetitive stimulation during protocol 1
was 11.5 ± 1.2 Hz. The pulse durations used for repetitive
stimulation during protocols 2 and 3 were 150 ± 22 and
131 ± 24 μs, respectively. Student’s paired t test confirmed
that the pulse durations used for the fatigue protocols
were significantly different (P < 0.05). The peak forces
produced in response to the fatiguing trains during the
three fatigue protocols for a representative subject are
shown in Fig. 3. It is notable that though the first fatiguing
trains of the three protocols produced similar peak forces
(Fig. 3A), the last fatiguing train of protocol 1 produced
the highest and the last fatiguing train of protocol 3
produced the lowest peak force for this subject (Fig. 3B).
The repeated measures ANOVA showed no significant
difference in the initial peak force produced by the three
fatigue protocols (F = 1.09; n.s.). The average peak forces
in response to the first fatiguing trains were 21.2 ± 1.7,
21.5 ± 2.2 and 22.3 ± 2.2% MVIC for protocols 1, 2 and 3,
respectively.
Force responses to the fatiguing trains
There were significant differences in percentage decline
in peak force (F = 30.08; P < 0.01) between the first
and last fatiguing trains among the three protocols
(Fig. 4). Protocol 1, consisting of fatiguing trains with long
(600 μs) pulse duration and low frequency, produced the
smallest percentage decline in peak force (31.3 ± 9.4%),
and protocol 3, consisting of fatiguing trains with high
frequency (60 Hz) and short pulse duration, produced


C 2006 The Authors. Journal compilation 
C 2006 The Physiological Society
972 T. Kesar and S. Binder-Macleod Exp Physiol 91.6 pp 967–976

A 1st Train Protocol #1 B Last Train Protocol #1

Protocol #2 Protocol #2
Protocol #3 Protocol #3
Force (N)

120 120

Force (N)
60 60

0 0
0 200 400 600 0 200 400 600 800
Time (ms) Time (ms)

C Fatiguing Trains Protocol #1

Protocol #2
180 Protocol #3
Peak Force (N)

Figure 3. Raw force profiles of the first (A)

and last fatiguing trains (B) of a
90
representative subject for the 3 fatigue
protocols
The figure also shows peak forces produced in
0 response to each fatiguing train during the
0 50 100 150 3 fatigue protocols for a representative subject
Contraction # (C).

the largest percentage decline in peak force (51.3 ± 7.5%; smallest decline in peak force (21.5 ± 9.5%), protocol 2
Fig. 4). produced an intermediate decline (27.4 ± 8.2%), and
protocol 3 produced the largest decline (46.1 ± 6.7%) in
peak force in response to the 60 Hz testing train at the same
Force responses of testing trains pulse duration as the fatiguing trains (F = 77.23; P < 0.01;
Fig. 5). Average peak forces produced in response to the
Average peak forces produced in response to the 60 Hz testing trains at the 600 μs pulse duration for
60 Hz testing trains at the same pulse duration as the protocols 1, 2 and 3, respectively, were 486.6 ± 160.9,
fatiguing trains for protocols 1, 2 and 3, respectively, 491.7 ± 142.4 and 489.2 ± 151.8 N (prefatigue) and
were 481.5 ± 164.2, 238.7 ± 69.2 and 221.1 ± 78.6 N 362.6 ± 101.4, 416.4 ± 112.0 and 376.3 ± 114.1 N
(prefatigue) and 367.5 ± 102.8, 170.4 ± 44.3 and (postfatigue). Interestingly, for the 60 Hz testing trains at
117.1 ± 36.8 N (postfatigue). Protocol 1 produced the 600 μs pulse duration, protocol 2, consisting of medium
frequency and medium pulse duration fatiguing trains,

Fatiguing Trains
75
*
*
% Decline in Peak Force

50
*

25

0
Protocol #1 Protocol #2 Protocol #3
Low-Freq
Low-Freq 30-Hz
Medium-Freq High-Freq
600-PD Medium-PD Short-PD

Figure 4. The percentage decline in peak force produced Figure 5. The percentage decline in peak forces between pre-
between the first and the last fatiguing trains during the and postfatigue 60 Hz testing trains at the same pulse duration
3 fatigue protocols as used for the fatiguing trains (left) and at 600 μs pulse
PD, pulse duration. ∗ Significant differences between protocols duration (right)
(P < 0.01). ∗ Significant difference between protocols (P < 0.01).


C 2006 The Authors. Journal compilation 
C 2006 The Physiological Society
Exp Physiol 91.6 pp 967–976 Muscle fatigue during repetitive stimulation 973

produced significantly smaller declines in peak force
(14.7 ± 7%) than protocol 1 (23.6 ± 8.3%) or protocol 3
(22.4 ± 10.3%; F = 6.40; P = 0.01; Fig. 5).
Average peak forces produced in response to the 20 Hz
testing trains at the pulse duration of the fatiguing trains
for protocols 1, 2 and 3, respectively, were 363.2 ± 117.7,
175.0 ± 55.1 and 162.6 ± 53.9 N (prefatigue) and
224.7 ± 68.4, 73.6 ± 24.7 and 41.9 ± 17.1 N (postfatigue).
Average peak forces produced in response to the 20 Hz
testing trains at the 600 μs pulse duration for protocols 1, 2
and 3, respectively, were 366.2 ± 116.5, 375.7 ± 111.5
and 362.6 ± 110.1 N (prefatigue) and 221.4 ± 68.9,
253.0 ± 72.6 and 230.9 ± 70.3 N (postfatigue). For testing
trains at the pulse duration of the fatiguing trains, the
two-way repeated measures ANOVA (protocol × fatigue)
showed significant effects of protocol (F = 32.00;
P < 0.01) and fatigue (F = 308.93; P < 0.01) on the
20 Hz:60 Hz peak force ratios (Fig. 6A). There was a
significant interaction between protocol and fatigue
(F = 35.32; P < 0.01). There were no significant
differences in the prefatigue 20 Hz:60 Hz peak force ratios
among the three protocols. The postfatigue 20 Hz:60 Hz
peak force ratios showed significant differences among
protocols (F = 50.7; P < 0.01). For testing trains at the
pulse duration of the fatiguing trains, protocol 1 showed
the largest postfatigue 20 Hz:60 Hz ratio (0.61 ± 0.07),
and protocol 3 showed the smallest ratio (0.37 ± 0.11;
Fig. 6A). For testing trains at 600 μs pulse duration, the
two-way ANOVA showed a significant effect of fatigue
(F = 66.95; P < 0.01), no significant effect of protocol
(F = 0.03; n.s.), and no significant interaction between the
effects of protocol and fatigue (F = 2.62; n.s.; Fig. 6B).
during muscle activation (Homsher, 1987; Baker et al.
1994; Frank et al. 1998; Hogan et al. 1998). Since protocol 1
used the fewest number of pulses and generated the
fewest action potentials, the least ATP utilization by the
Ca2+ -ATPase and Na+ –K+ -ATPase reactions also occurred
during protocol 1 (Marsden et al. 1983; Frank et al. 1998).
Since metabolic demand is related to muscle fatigue
(Cooke et al. 1988; Sahlin et al. 1998; Westerblad et al.
1998), we believe that protocol 1 produced the least fatigue
in the motor unit population recruited by the fatiguing
trains because the lowest ATP consumption and lowest
metabolic demand was placed on each muscle fibre during
protocol 1 among the three protocols tested.
Recently, Godfrey et al. (2002) studied the effects of
stimulation intensity on force production of thenar hand
muscles in patients with spinal cord injury. In contrast
to our present findings, Godfrey et al. (2002) found
greater fatigue during stimulation at supra- compared to
submaximal intensities. We believe that the differences
in the findings of Godfrey et al. (2002) versus our
present study resulted from methodological differences.
Since Godfrey et al. (2002) delivered both supra- and
submaximal stimulation intensities at the same frequency
(40 Hz), different force levels were generated at the two
intensities. The higher forces generated at supra- versus

Discussion
Our study compared the performance and muscle
fatigue produced during repetitive stimulation using three
different combinations of frequencies and pulse durations,
and found that protocol 1 produced the least muscle
fatigue in the motor units recruited by the fatiguing trains
(see Fig. 5). Of the three protocols tested, protocol 1 used
the lowest frequency (11.5 ± 1.2 Hz) and highest level of
recruitment (600 μs pulse duration; Robinson & Snyder-
Mackler, 1995) to fatigue the muscle. Since protocol 1
recruited the most motor units, and all three protocols
generated the same target peak force, the least amount
of force was generated by each active muscle fibre during
protocol 1. Since ATP utilization by actin–myosin ATPase Figure 6. Ratios of peak forces produced in response to the pre-
is proportional to force generation (Boska, 1994; Potma and postfatigue 20 versus 60 Hz testing trains (20 Hz:60 Hz peak
et al. 1994), we suggest that protocol 1 resulted in the force ratio) at the same pulse duration as used for the fatiguing
least ATP utilization by the actin–myosin ATPase per trains (A) and at the 600 μs pulse duration (B) during the
3 fatigue protocols
muscle fibre. In addition to the actin–myosin ATPase, the ∗ Significant differences between fatigue protocols (P < 0.05). All
Ca2+ -ATPase and Na+ –K+ -ATPase reactions in response prefatigue 20 Hz:60 Hz peak force ratios were significantly different
to each action potential contribute to ATP utilization from the postfatigue ratios (P < 0.05).


C 2006 The Authors. Journal compilation 
C 2006 The Physiological Society
974 T. Kesar and S. Binder-Macleod
submaximal intensities probably contributed to the greater
muscle fatigue observed by Godfrey et al. (2002) at the
supramaximal intensities (Russ et al. 2002b,c). However,
the differences between the results of our present study
and those of Godfrey et al. (2002) may also result from
differences in the subject populations and the muscle
tested.
Protocol 1 also produced the least low-frequency
fatigue in the motor units recruited by the fatiguing
trains (Fig. 6A). Low-frequency fatigue is the result of
impairment in excitation–contraction coupling that is
thought to result from increased levels of intracellular
Ca2+ during muscle activation (Westerblad et al. 1993;
Chin & Allen, 1996; Chin et al. 1997). Intracellular Ca2+
concentrations have been shown to be directly related to
the stimulation frequency (Chin & Allen, 1996). The low
frequency used during protocol 1 would therefore result
in the lowest levels of intracellular Ca2+ and the least low-
frequency fatigue among the three protocols (Westerblad
et al. 1993; Chin & Allen, 1996; Chin et al. 1997).
In contrast to our present findings, Matsunaga et al.
(1999) showed smaller declines in force during repetitive
stimulation at high- (100 Hz) versus low-frequency
(20 Hz) activation. Compared to our study, the fatigue
protocols tested by Matsunaga et al. (1999) were of much
longer durations (60 min versus 180 s in our study), used
much shorter duty cycles (1:15, 1:30 and 1:60 versus 1:2.3
in our study), and produced smaller percentage declines
in peak force (22.3 ± 15.1% at 100 Hz versus 51.3 ± 7.5%
at 60 Hz in our study). Since muscle performance is a
function of the extent of force fatigue and low-frequency
fatigue produced by the fatiguing trains in the motor unit
population recruited by the fatiguing trains, we believe that
the results of Matsunaga et al. (1999) can be explained by
greater low-frequency fatigue and less muscle fatigue than
presently observed. That is, even if greater muscle fatigue
was produced by the 100 Hz trains than the 20 Hz trains in
their study (Matsunaga et al. 1999), low-frequency fatigue
may have markedly reduced the muscles’ responses to the
20 Hz trains and therefore resulted in poorer performance
in response to the 20 Hz trains.
The frequencies used during protocols 1, 2 and 3 in our
study were on the low, middle and high ranges of the force–
frequency curves, respectively. Surprisingly, however,
the average pulse durations used during protocols 2
(150 ± 22 μs) and 3 (131 ± 24 μs), although significantly
different, only varied by 19 μs. A possible reason could
be that the pulse durations used for protocols 2 and 3
were in the steep rising part of the force versus pulse
durations curves. Thus, our results showed that although
the difference in frequencies between protocols 2 and 3
was relatively large (30 versus 60 Hz), a relatively small
difference in pulse duration between protocol 3 and
protocol 2 enabled the targeted 20% MVIC force to be
reached for both the protocols.
Exp Physiol 91.6 pp 967–976
The present findings could have implications for the
development of strategies for optimal activation of skeletal
muscle during FES. Functional electrical stimulation is
used to generate functional movements in patients with
upper motor neurone lesions, such as spinal cord injury,
hemiplegia following stroke, and cerebral palsy. This
study represents the first step in a project whose long-
term aim is to develop electrical stimulation strategies
that can maximize FES performance. Initial testing on
healthy individuals has helped us to identify hypotheses
that can then be tested on paralysed muscles using fewer
experimental sessions. Interestingly, consistent with out
present findings, recent studies showed that for healthy
subjects and for subjects with spinal cord injury, starting
at low frequencies and later switching to high frequencies
produced better performance during repetitive non-
isometric contractions than stimulation using either a
low or high frequency alone (Kebaetse & Binder-Macleod,
2004; Kebaetse et al. 2005). Starting repetitive stimulation
with low frequencies produced less muscle fatigue, and
switching to a higher stimulation frequency allowed the
stimulation to overcome the effects of low-frequency
fatigue (Kebaetse & Binder-Macleod, 2004; Kebaetse et al.
2005). Future studies will need to identify the best
frequency and intensity of the initial trains, as well as the
strategies for modulation of frequency and intensity of
the subsequent trains, to maximize muscle performance
during FES.
An interesting finding of this study was the difference
in the responses to the testing trains when the pulse
duration was maintained at the levels used to fatigue
the muscle versus when a 600 μs pulse duration was
used (see Figs 5 and 6). During protocols 2 and 3, the
postfatigue testing trains at 600 μs pulse duration activated
previously unrecruited motor units. The responses to the
testing trains at 600 μs pulse duration were therefore
the sum of the forces produced by the recruited and
previously unrecruited motor units. Protocol 3 showed
the most and protocol 1 the least muscle fatigue and low-
frequency fatigue when the pulse duration was maintained
at the level used to fatigue the muscle (Fig. 5). Thus,
during FES applications, if the frequency and pulse
duration are held constant during repetitive stimulation,
using the lowest frequency and longest pulse duration
may maximize performance. However, Protocols 3 and
1 showed comparable amounts of muscle fatigue and
protocol 2 showed the least muscle fatigue when tested at
the 600 μs pulse duration (Fig. 5). In addition, in response
to testing trains at 600 μs pulse duration, there were no
differences in the overall levels of low-frequency fatigue
among the three protocols (Fig. 6B). This is an important
finding because most FES systems used at present deliver
a constant frequency and increase the intensity to increase
muscle force output as the muscle fatigues (Weingarden
et al. 1997; Taylor et al. 1999; Weber et al. 2005). Thus,

C 2006 The Authors. Journal compilation 
C 2006 The Physiological Society
Exp Physiol 91.6 pp 967–976 Muscle fatigue during repetitive stimulation
for FES applications where intensity is increased as the
muscle fatigues, a ‘medium’ frequency, similar to the
frequency used in protocol 2 of our study, may minimize
muscle fatigue. The combination of frequency and pulse
duration that minimizes muscle fatigue and/or maximizes
performance during FES may depend on whether or not
modulation of frequency or intensity will be used as
strategies during repetitive stimulation.
Conclusion
The present findings support the hypothesis that when
the same initial peak force is generated using different
combinations of frequency and pulse duration, and when
the frequency and pulse duration are kept constant
throughout repetitive stimulation, repetitive stimulation
with a long pulse duration (600 μs) and low frequency
(11.5 ± 1.2 Hz; protocol 1) would maximize isometric
performance by minimizing muscle fatigue. Interestingly,
repetitive stimulation with a medium frequency (30 Hz)
and medium pulse duration (150 ± 22 μs; protocol 2)
produced the least muscle fatigue when comparable
motor unit populations were tested across protocols
(i.e. at the 600 μs pulse duration). These findings may
have important implications for the design of optimal
stimulation strategies to use during FES. Specifically, the
present results should help in the design of future studies
involving patient populations where complex stimulation
strategies that modulate both the frequency and pulse
durations will be tested.
References
Agarwal S, Kobetic R, Nandurkar S & Marsolais EB (2003).
Functional electrical stimulation for walking in paraplegia:
17-year follow-up of 2 cases. J Spinal Cord Med 26, 86–91.
Baker AJ, Brandes R, Schendel TM, Trocha SD, Miller RG &
Weiner MW (1994). Energy use by contractile and
noncontractile processes in skeletal muscle estimated by
31
P-NMR. Am J Physiol 266, C825–C831.
Bigland-Ritchie B, Jones DA & Woods JJ (1979). Excitation
frequency and muscle fatigue: electrical responses during
human voluntary and stimulated contractions. Exp Neurol
64, 414–427.
Binder-Macleod SA, Dean JC & Ding J (2002). Electrical
stimulation factors in potentiation of human quadriceps
femoris. Muscle Nerve 25, 271–279.
Binder-Macleod SA, Halden EE & Jungles KA (1995). Effects of
stimulation intensity on the physiological responses of
human motor units. Med Sci Sports Exerc 27, 556–565.
Binder-Macleod SA, Lee SC, Russ DW & Kucharski LJ (1998).
Effects of activation pattern on human skeletal muscle
fatigue. Muscle Nerve 21, 1145–1152.
Binder-Macleod SA & Snyder-Mackler L (1993). Muscle
fatigue: clinical implications for fatigue assessment and
neuromuscular electrical stimulation. Phys Ther 73,
902–910.

C 2006 The Authors. Journal compilation 
C 2006 The Physiological Society
975
Boska M (1994). ATP production rates as a function of force
level in the human gastrocnemius/soleus using 31 P MRS.
Magn Reson Med 32, 1–10.
Chin ER & Allen DG (1996). The role of elevations in
intracellular [Ca2+ ] in the development of low frequency
fatigue in mouse single muscle fibres. J Physiol 491, 813–824.
Chin ER, Balnave CD & Allen DG (1997). Role of intracellular
calcium and metabolites in low-frequency fatigue of mouse
skeletal muscle. Am J Physiol 272, C550–C559.
Cooke R, Franks K, Luciani GB & Pate E (1988). The inhibition
of rabbit skeletal muscle contraction by hydrogen ions and
phosphate. J Physiol 395, 77–97.
Davoodi R, Andrews BJ, Wheeler GD & Lederer R (2002).
Development of an indoor rowing machine with manual FES
controller for total body exercise in paraplegia. IEEE Trans
Neural Syst Rehabil Eng 10, 197–203.
Donaldson N, Perkins TA, Fitzwater R, Wood DE & Middleton
F (2000). FES cycling may promote recovery of leg function
after incomplete spinal cord injury. Spinal Cord 38, 680–682.
Edwards RH (1981). Human muscle function and fatigue. Ciba
Found Symp 82, 1–18.
Frank K, Bolck B, Bavendiek U & Schwinger RH (1998).
Frequency dependent force generation correlates with
sarcoplasmic calcium ATPase activity in human
myocardium. Basic Res Cardiol 93, 405–411.
Garland SJ, Garner SH & McComas AJ (1988). Relationship
between numbers and frequencies of stimuli in human
muscle fatigue. J Appl Physiol 65, 89–93.
Gerrits HL, De Haan A, Hopman MT, van Der Woude LH,
Jones DA & Sargeant AJ (1999). Contractile properties of the
quadriceps muscle in individuals with spinal cord injury.
Muscle Nerve 22, 1249–1256.
Gerrits HL, Hopman MT, Offringa C, Engelen BG, Sargeant
AJ, Jones DA & Haan A (2003). Variability in fibre properties
in paralysed human quadriceps muscles and effects of
training. Pflugers Arch 445, 734–740.
Gfohler M & Lugner P (2000). Cycling by means of functional
electrical stimulation. IEEE Trans Rehabil Eng 8, 233–243.
Godfrey S, Butler JE, Griffin L & Thomas CK (2002).
Differential fatigue of paralyzed thenar muscles by stimuli of
different intensities. Muscle Nerve 26, 122–131.
Grill WM Jr & Mortimer JT (1996). The effect of stimulus pulse
duration on selectivity of neural stimulation. IEEE Trans
Biomed Eng 43, 161–166.
Hogan MC, Ingham E & Kurdak SS (1998). Contraction
duration affects metabolic energy cost and fatigue in skeletal
muscle. Am J Physiol 274, E397–E402.
Homsher E (1987). Muscle enthalpy production and its
relationship to actomyosin ATPase. Annu Rev Physiol 49,
673–690.
Hunt KJ, Stone B, Negard NO, Schauer T, Fraser MH, Cathcart
AJ, Ferrario C, Ward SA & Grant S (2004). Control strategies
for integration of electric motor assist and functional
electrical stimulation in paraplegic cycling: utility for
exercise testing and mobile cycling. IEEE Trans Neural Syst
Rehabil Eng 12, 89–101.
Johnston TE, Betz RR, Smith BT, Benda BJ, Mulcahey MJ,
Davis R, Houdayer TP, Pontari MA, Barriskill A & Creasey
GH (2005). Implantable FES system for upright mobility and
bladder and bowel function for individuals with spinal cord
injury. Spinal Cord 43, 713–723.
976 T. Kesar and S. Binder-Macleod
Kebaetse MB & Binder-Macleod SA (2004). Strategies that
improve human skeletal muscle performance during
repetitive, non-isometric contractions. Pflugers Arch 448,
525–532.
Kebaetse MB, Lee SC, Johnston TE & Binder-Macleod SA
(2005). Strategies that improve paralyzed human quadriceps
femoris muscle performance during repetitive, nonisometric
contractions. Arch Phys Med Rehabil 86, 2157–2164.
Kralj A, Bajd T & Turk R (1988). Enhancement of gait
restoration in spinal injured patients by functional electrical
stimulation. Clin Orthop Relat Res 233, 34–43.
Liberson WT, Holmquest HJ, Scot D & Dow M (1961).
Functional electrotherapy: stimulation of the peroneal nerve
synchronized with the swing phase of the gait of hemiplegic
patients. Arch Phys Med Rehabil 42, 101–105.
Mangold S, Keller T, Curt A & Dietz V (2005). Transcutaneous
functional electrical stimulation for grasping in subjects with
cervical spinal cord injury. Spinal Cord 43, 1–13.
Marsden CD, Meadows JC & Merton PA (1983). ‘Muscular
wisdom’ that minimizes fatigue during prolonged effort in
man: peak rates of motoneuron discharge and slowing of
discharge during fatigue. Adv Neurol 39, 169–211.
Marsolais EB & Edwards BG (1988). Energy costs of walking
and standing with functional neuromuscular stimulation
and long leg braces. Arch Phys Med Rehabil 69, 243–249.
Matsunaga T, Shimada Y & Sato K (1999). Muscle fatigue from
intermittent stimulation with low and high frequency
electrical pulses. Arch Phys Med Rehabil 80, 48–53.
Peckham PH & Knutson JS (2005). Functional electrical
stimulation for neuromuscular applications. Annu Rev
Biomed Eng 7, 327–360.
Pierrynowski MR & Morrison JB (1985). Estimating the muscle
forces generated in the human lower extremity when
walking: a physiological solution. Math Biosci 75,
43–68.
Potma EJ, Stienen GJ, Barends JP & Elzinga G (1994).
Myofibrillar ATPase activity and mechanical performance of
skinned fibres from rabbit psoas muscle. J Physiol 474,
303–317.
Riener R (1999). Model-based development of neuroprosthesis
for paraplegic patients. Philos Trans R Soc Lond B Biol Sci
354, 877–894.
Ring H & Rosenthal N (2005). Controlled study of
neuroprosthetic functional electrical stimulation in
sub-acute post-stroke rehabilitation. J Rehabil Med 37,
32–36.
Robinson AJ & Snyder-Mackler L (1995). Clinical
Electrophysiology, Electrotherapy and Electrophysiologic
Testing. Williams & Wilkins, Baltimore.
Russ DW & Binder-Macleod SA (1999). Variable-frequency
trains offset low-frequency fatigue in human skeletal muscle.
Muscle Nerve 22, 874–882.
Russ DW, Elliott MA, Vandenborne K, Walter GA &
Binder-Macleod SA (2002a). Metabolic costs of force
generation and maintenance of human skeletal muscle.
Am J Physiol Endocrinol Metab 282, E448–E457.
Exp Physiol 91.6 pp 967–976
Russ DW, Vandenborne K & Binder-Macleod SA (2002b).
Factors in fatigue during intermittent electrical stimulation
of human skeletal muscle. J Appl Physiol 93, 469–478.
Russ DW, Vandenborne K, Walter GA, Elliott M &
Binder-Macleod SA (2002c). Effects of muscle activation on
fatigue and metabolism in human skeletal muscle. J Appl
Physiol 92, 1978–1986.
Sahlin K, Tonkonogi M & Soderlund K (1998). Energy supply
and muscle fatigue in humans. Acta Physiol Scand 162,
261–266.
Snyder-Mackler L, De Luca PF, Williams PR, Eastlack ME &
Bartolozzi AR 3rd (1994). Reflex inhibition of the quadriceps
femoris muscle after injury or reconstruction of the anterior
cruciate ligament. J Bone Joint Surg Am 76, 555–560.
Taylor PN, Burridge JH, Dunkerley AL, Wood DE, Norton JA,
Singleton C & Swain ID (1999). Clinical use of the Odstock
dropped foot stimulator: its effect on the speed and effort of
walking. Arch Phys Med Rehabil 80, 1577–1583.
Triolo RJ, Bieri C, Uhlir J, Kobetic R, Scheiner A & Marsolais
EB (1996). Implanted functional neuromuscular stimulation
systems for individuals with cervical spinal cord injuries:
clinical case reports. Arch Phys Med Rehabil 77, 1119–1128.
Vollestad NK (1997). Measurement of human muscle fatigue.
J Neurosci Meth 74, 219–227.
Weber DJ, Stein RB, Chan KM, Loeb G, Richmond F, Rolf R,
James K & Chong SL (2005). BIONic WalkAide for correcting
foot drop. IEEE Trans Neural Syst Rehabil Eng 13, 242–246.
Weingarden HP, Kizony R, Nathan R, Ohry A & Levy H (1997).
Upper limb functional electrical stimulation for walker
ambulation in hemiplegia: a case report. Am J Phys Med
Rehabil 76, 63–67.
Westerblad H, Allen DG, Bruton JD, Andrade FH &
Lannergren J (1998). Mechanisms underlying the reduction
of isometric force in skeletal muscle fatigue. Acta Physiol
Scand 162, 253–260.
Westerblad H, Duty S & Allen DG (1993). Intracellular calcium
concentration during low-frequency fatigue in isolated single
fibers of mouse skeletal muscle. J Appl Physiol 75, 382–388.
Wheeler GD, Andrews B, Lederer R, Davoodi R, Natho K,
Weiss C, Jeon J, Bhambhani Y & Steadward RD (2002).
Functional electric stimulation-assisted rowing: increasing
cardiovascular fitness through functional electric stimulation
rowing training in persons with spinal cord injury. Arch Phys
Med Rehabil 83, 1093–1099.
Acknowledgements
The authors would like to thank R. Perumal and R. Maladen
for their helpful comments on an early draft of this manuscript
and for the development of the software and hardware for data
acquisition. Also, we thank Li-Wei Chou for his assistance and
suggestions for the design of the study. This work was supported
by NIH grants no. HD36797 and no. HD38582.

C 2006 The Authors. Journal compilation 
C 2006 The Physiological Society