This study examined the effects of arm training on retention of training effects produced by leg training. Thirteen adult male subjects leg trained on bicycle ergometers for 8 weeks. During the subsequent 4 week control period group AT (N = 4) arm trained by cranking an ergometer, group DT (N = 4) discontinued training and group LT (N = 5) continued leg training. Metabolic response to maximal and submaximal (80% initial Vo2max) leg work was observed before (Test I) and after (Test II) the 8 week leg training program and at the conclusion of the control period (Test III). With the initial 8 weeks of leg training Vo2max and maximal Ve increased while Vo2, HR, Ve and blood lactate decreased during submaximal work (p less than .05). R was decreased during both maximal and submaximal work (p less than .05). Analysis of covariance of the Test III means (Test II as covariate) yielded significant (p less than .05)F-ratios for Vo2max and R at Vo2max. Post hoc testing indicated that Vo2max was greater in LT than in AT or DT but that there was no difference between AT and DT. R at Vo2max was greater in DT than in LT or AT. These data suggest that arm training does not significantly affect the deterioration in metabolic response to leg work which occurs with cessation of leg training.