Fifteen children with autism were treated with 60 mg d,l-fenfluramine (FEN) or placebo in a double-blind A-B-A protocol followed immediately by double-blind placebo-controlled crossover administration of FEN (total duration 62 weeks). Both biochemical and clinical outcomes were examined. Biochemically, FEN led to an increase in dihydroxyphenylacetic acid (DOPAC) and decreases in whole-blood serotonin (5-HT), plasma norepinephrine (NE), and plasma 3-methoxy-4-hydroxyphenylglycol (MHPG). The decrease in whole-blood 5-HT was seen only during treatment with FEN. However, NE levels did not return to baseline as long as 8 weeks after the first FEN treatment period. Increases in DOPAC were greater during the second FEN treatment period than the first. Persistent changes in catecholamine regulation may be related to previously reported long-term effects on central nervous system 5-HT after FEN. Clinically, FEN led to a modest decrease in parent, but not teacher, ratings of hyperactivity and to a small reduction in sensorimotor abnormalities. Abnormal social and affectual responses also decreased, but this was not directly related to FEN treatment. Effects on cognition were equivocal. Hyperserotonemic subjects did not differ from normoserotonemic subjects in clinical response. Overall, no significant advantage for the use of FEN could be established.