The purpose of this study was to compare the electromyographic fatigue threshold (EMG(FT)) values determined simultaneously from superficial elbow flexor muscles during an isometric single-joint task. Eight subjects performed isometric elbow flexions at randomly ordered percentages of maximal voluntary contraction (20, 30, 40, 50 and 60%). During these bouts, electromyographic (EMG) activity was measured in the anterior head of Deltoïd, lateral head of Triceps brachii, Brachioradialis and both short and long head of Biceps brachii. For each subject and each muscle, the EMG amplitude data were plotted as function of time for the five submaximal bouts. The slope coefficient of the EMG amplitude versus time linear relationships were plotted against force level. EMG(FT) was determined as the y-intercept of this relationship and considered as valid only if the following criteria were met: (1) significant positive linear regression (P < 0.05) between force and slope coefficient, (2) an adjusted coefficient of determination for force versus slope coefficient relationship greater than 0.85, and (3) a standard error for the EMG(FT) below 5% of maximal voluntary contraction. The EMG(FT) could only be determined for one muscle (the long head of Biceps brachii) and only in three out of the eight subjects (mean value = 24.9 +/- 1.1% of maximal voluntary contraction). The lack of EMG(FT) in most of the subjects (5/8) could be explained by putative compensations between elbow muscles which were indirectly observed in some subjects. In this way, EMG(FT) should be studied from a more simple movement i.e., ideally a movement implying mainly one muscle.