Double-blind crossover comparison of methylphenidate hydrochloride, dextroamphetamine sulfate, and caffeine after placebo washout in 29 children with minimal brain dysfunction (MBD) showed on six ratings that methylphenidate and dextroamphetamine were significantly (P less than .05 to P less than .001) better than placebo and caffeine, but not significantly (P less than .05) different from each other. Placebo, caffeine, and ratings before drug did not differ significantly. Of 26 drug responders, 12 responded best to dextroamphetamine, ten to methylphenidate, and one to caffeine. The latter child showed no improvement at all with either prescription stimulant. Methylphenidate and dextroamphetamine were each efficacious for six children who did not respond to the other stimulant. All three drugs showed significant (P less than .05) weight loss and cardiovascular side effects, the latter possibly spurious. Dextroamphetamine showed a significant (P less than .05) decrease from placebo in "tummyaches."