Maximal anaerobic capacity, i.e., the maximal amount of energy released by anaerobic processes (E(an max), J.kg-1), has been evaluated from maximal increase of plasma lactate concentration (Lap) in eight male subjects of different physical fitness submitted to superamaximal runs of various intensity performed until volitional exhaustion (temps-limite, t(lim)). As previously found (2), the interindividual differences of t(lim) were reduced when exercise intensity was expressed by the anaerobic component of exercise defined as the difference between the overall energy requirement (E, W.kg-1) and maximal aerobic power E(ox max), W.kg-1). Within the range of intensity studied, Lap did not vary significantly as a function of E-E(ox max). However, the performances achieved by the less fit subjects (group 1) remained lower than those achieved by the more athletic subjects (group 2). Mean Lap were significantly higher in group 2 (17.2 mmol.l-1) than in group 1 (13.7 mmol.l-1). The rate of increase of Lap, defined by the ratio Lap/t(lim), was a linear function of E-E(ox max). The energy equivalent of plasma lactate accumulation (beta), given by the reciprocal of the slope of the equation describing the relationship Lap/t(lim) = f(E-E(ox max)), amounts 56.8 J.kg-1 when Lap is increased by 1 mmol.l-1. The energy released by anaerobic glycolysis was calculated by multiplying beta by mean Lap measured in the two groups of subjects. Assuming that the energy yielded by the anaerobic alactic processes amounts 260 J.kg-1(1), mean E(an max) values in group 1 and 2 were found to be equal to 1040 (range: 910-1110) and 1240 J.kg-1 (range: 1100-1330), respectively. In order to validate these results, we developed a model relating t(lim) to E(an max), E(ox max) and the overall energy cost of exercise on the basis of the energy conservation principle. As the theoretical relationships t(lim) = f(E-E(ox max)) derived from our model fitted the experimental results quite well, we concluded that E(an max) has been correctly evaluated.