The purpose of this study was to evaluate the test for estimating physical working capacity at fatigue threshold. The subjects in this study were 20 female college students including highly trained endurance athletes (TR:n=6) and untrained sedentary individuals (UTR:n=6). The fatigue threshold was measured by recording electromyogram (EMG) fatigue curves from the quadriceps muscle in a series of discontinuous, work bouts (200, 250, 300, 350 W) on an electrically-braked bicycle ergometer. The power output at the fatigue threshold (EMG_<FT>) was defined as the greatest power loading without electromyographic evidence of muscle fatigue, i.e.the highest constant power output of the subject on the bicycle ergometer which resulted no increase in integrated EMG (iEMG) as a function of time (iEMG slope). Anaerobic threshold (AT)was estimated from gas exchange parameters during an incremental work test on the same bicycle ergometer. The results indicated that 1) AT VO^^._2 (1.39 ± 0.44 l・min^<-1>) and EMG_<FT> VO_2 (1.33 ± 0.57 l・min^<-1>) measured for a entire group of subjects were correlated at r=0.823 with no significant difference (p>0.05) between these mean values. 2) Statistically significant differences were found between TR and UTR for VO^^._2max, AT VO^^._2 and EMG_<FT> VO^^._2 (p<0.01). These results suggested the validity and feasibility of EMG_<FT> determination using iEMG slope coefficient so obtained during fatiguing exercise on the bicycle ergometer. Since this method could be achieved by several discontinuous maximum 1 minute work bouts, it appears to be reasonably simple while minimizing the psychological factors affecting the results.