The determination of anaerobic capacity (AC) using treadmill ergometry is problematic from a methodological, as well as a technical standpoint. In this study, a procedure from Monod and Scherrer was modified to examine whether realistic magnitudes of AC could be determined using three subject groups with different levels of anaerobic training. The subject groups consisted of 10 untrained (UT), 10 aerobic-trained runners (AeT), and 10 anaerobic-trained 400-meter sprinters (AnT). In two separate test series, first the VO2max was determined and second the so-called individual anaerobic threshold (IAT) was used to determine the aerobic power for all subjects. Then all subjects completed a series of sprints with increasing speeds above the VO2max, from which the work output from each test was calculated. Through linear regression, the point of intersection of the regression line with the y-axis was defined as global AC. The results show typically higher VO2max and IAT for AeT (62.2 ml x kg(-1) x min(-1), 14.7 km x h(-1)) compared to UT (53.2 ml x kg(-1) x min(-1); 11.2 km x h(-1)) and AnT (56.7 ml x kg(-1) x min(-1); 11.8 km x h(-1)). AC was significantly higher in AnT (4.1 +/- 0.58 kJ) compared to AeT (1.8 +/- 0.65 kJ) and UT (3.2 +/- 0.68 kJ). The determined absolute values of AC are considerably lower than of comparable examinations using bicycle ergometry. One reason for such an underestimation of AC could be that the horizontal work done during exercise on a treadmill was not taken into enough consideration. Another explanation is that the magnitude of the calculated AC values shows a dependency on the duration of each sprint test. In addition, the critical velocity for all subjects was found to be higher than for IAT, which consequently leads to an underestimation of AC. Moreover, the absolute level of the AC values appears to depend on the endurance of the comparison groups. It can then be concluded that the applied procedure allows for a differentiation amongst a variously trained collective, but does not allow a correct absolute determination of the AC.