In infantile autism, the serotoninergic (5-HT) hypothesis is corroborated by biological dosages and therapeutic effects of fenfluramine which decrease blood serotonin. However other drugs, such as dopaminergic agonists or antagonists, have therapeutic effects. Therefore, we tested the hypothesis that two dopaminergic (DA) drugs have a similar 5-HT effect underlying the therapeutic efficiency. We evaluated in a randomized, double-blind and cross-over study, the effects of a DA agonist (bromocriptine) and a DA antagonist (amisulpride) on platelet 5-HT in infantile autism. The prolactinemia, reflecting the DA action, has been also measured. Nine children, aged from 4 to 13 years, according to the DSM III for infantile autism, received either drug in a random order during four weeks with an in-between placebo period of six weeks. The dosages of platelet 5-HT and serum prolactin were carried out at the beginning and at the end of every phase of treatment (active or placebo) with radioenzymology and radioimmunoassay methods respectively. The principal results on serum prolactin show neither order x treatment interaction, nor order effect but a significant treatment effect (p < 0.01): amisulpride increases serum prolactin whereas bromocriptine decreases according to the usual data. About platelet 5-HT, there is neither order x treatment interaction, nor treatment effect but a significant order effect (p < 0.01). Both drugs increase platelet 5-HT in the first phase of treatment. This order effect could be explained by a remanent effect of amisulpride after 6 wash-out weeks.(ABSTRACT TRUNCATED AT 250 WORDS)