For this study, Baumeister's variability hypothesis (1968) was investigated in eight 10-year-old mildly mentally retarded (MMR) and 16 nonhandicapped boys matched on either mental age (MA) or chronological age (CA). The boys attempted 60 trials on a novel kicking task. Results indicated that CA-matched boys exhibited less absolute error than either MMR- or MA-matched boys, while the latter two groups performed similarly. Variable error was observed to be significantly greater for MA-matched boys than for either MMR- or CA-matched boys, while MMR boys performed with greater although not significant variable error than CA-matched boys. For this study, Baumeister's variability hypothesis (1968) was investigated in eight 10-year-old mildly mentally retarded (MMR) and 16 nonhandicapped boys matched on either mental age (MA) or chronological age (CA). The boys attempted 60 trials on a novel kicking task. Results indicated that CA-matched boys exhibited less absolute error than either MMR- or MA-matched boys, while the latter two groups performed similarly. Variable error was observed to be significantly greater for MA-matched boys than for either MMR- or CA-matched boys, while MMR boys performed with greater although not significant variable error than CA-matched boys