In order to validate a maximal multistage 20-m shuttle run test for the prediction of VO2 max, 91 adults (32 females and 59 males, aged 27.3 +/- 9.2 and 24.8 +/- 5.5 year respectively and with mean VO2 max (+/- SD) of 39.3 +/- 8.3 and 51.6 +/- 7.8 ml . kg-1 . min-1 respectively) performed the test and had VO2 max estimated by the retroextrapolation method (extrapolation to time zero of recovery of the exponential least squares regression of the first four 20-s recovery VO2 values). Starting at 8 km . h-1 and increasing by 0.5 km . h-1 every 2 min, the 20-m shuttle run test enabled prediction of the VO2 max (y, ml . kg-1 . min-1) from the maximal speed (x, km . h-1) by means of the following regression equation: y = 5.857x - 19.458; r = 0.84 and SEE = 5.4. Later, the multistage protocol was slightly modified to its final version, in which the test started at stage 7 Met and continued with a 1 Met (3.5 ml O2 . kg-1 . min-1) increment every 2 min. Twenty-five of the 91 subjects performed the 20-m shuttle test twice, once on a hard, low-friction surface (vinyl-asbestos tiles) and another time on a rubber floor, as well as a walking maximal multistage test on an inclined treadmill. There was no difference between the means of these tests or between the slopes of the VO2max - maximal speed regressions for the two types of surfaces. The 20-m shuttle run test and another maximal multistage field test involving continuous track running gave comparable results (r = 0.92, SEE = 2.6 ml O2 . kg-1 . min-1, n = 70). Finally, test and retest of the 20-m shuttle run test also yielded comparable results (r = 0.975, SEE = 2.0 ml O2 . kg-1 . min-1, n = 50). It is concluded that the 20-m shuttle run test is valid and reliable test for the prediction of the VO2 max of male and female adults, individually or in groups, on most gymnasium surfaces.