The maximal oxygen uptake (VO2max) of 13 swimmers was determined by an arm-ergometer test (direct method) and estimated from a maximal multistage swimming test (indirect method) (23). A test-retest of the progressive swimming exercise showed that there were no significant differences from one test to the other and that there were significant correlations between the principal parameters: arm stroke index: 0.73, maximal aerobic swimming velocity: 0.94, VO2max: 0.95, p less than 0.01. Therefore, for swimmers of average ability, the reproducibility of this test has been proved. A significant difference (p less than 0.001) was observed between the two tests for VO2max: arm-ergometer test (VO2max arms): 2.4 +/- 0.5 l.min-1, swimming test (VO2max ST): 3.2 +/- 0.7 l.min-1, p less than 0.01. This difference appeared to be linked to the use of a greater muscle mass (arms and legs) during swimming. A significant correlation (r = 0.73, p less than 0.01) was obtained between VO2max (l.min-1) by using both the direct and indirect exercises as methods of measurement. However, the level of r did not permit the prediction of one parameter from the other. Significant correlations were obtained between VO2max and performances over 200 and 400 m free style regardless of the methodology used (VO2max arm, VO2max ST). Moreover, only VO2max (arm, ST) emerged as a variable accounting for swimming performance from a step-wise multiple regression analysis, in which biometric and bioenergetic parameters were taken into account.