The purpose of this study was to determine if a step test could be feasible, reliable, and valid for youths with educable (EMR) or trainable (TMR) mental retardation. Thirteen males and 11 females (age M = 14.7 ± 2.7 yr) with EMR or TMR participated in this study. Three step tests were employed using one platform height and stepping frequencies of 13, 15, and 17 ascents/min for 3 min. Recovery HR was used to estimate VO2peak. Though significant, correlations between the recovery HR and VO2peak for the 15 (r = −0.48) and 17 (r = −0.46) ascents/min were not high enough to be considered valid indicators of VO2peak. The large standard errors of the estimate and total errors suggested systematic errors of prediction. Furthermore, the measured VO2peak was significantly different from the estimated values at all step rates (p < .05). The step-test was relatively feasible, but was not a valid test of VO2peak in this population.