Double-blind crossover randomized Latin square comparison of placebo, dextroamphetamine, and levoamphetamine in 31 consecutively diagnosed children with minimal brain dysfunction (MBD) replicated a smaller nonrandom study. Both isomers showed significantly more benefit than placebo but were not significantly different from each other. Dextroamphetamine showed a nonsignificant trend of superiority over levoamphetamine. Of 25 subjects who responded well to drugs, three responded only to levoamphetamine, five only to dextroamphetamine, and 17 to both. This study seems to confirm the efficacy of levoamphetamine in MBD. An unsocialized aggressive subgroup (308.4) showed a nonsignificant trend for levoamphetamine superiority, in contrast to the hyperkinetic (308.0) and overanxious (308.2) subgroups. Those who responded best to levoamphetamine tended (not significantly) to be from poorer functioning families. Parents' ratings, but not teachers' or psychiatrists' ratings, showed significant placebo effect.