Valproic acid (VPA) may cause impaired platelet function, thrombocytopenia and, occasionally, severe bleeding. Controversy exists both as to the mechanism of this alteration in hemostasis and as to whether these adverse effects are either dose-related or idiosyncratic. Previous investigations have focused primarily on pediatric patients who commonly were receiving multiple anticonvulsant medications. We evaluated a cohort of 27 adult patients with epilepsy who were receiving VPA monotherapy and compared them with age-matched controls to determine whether a correlation exists between platelet count, function, bleeding time, levels of von Willebrand's factor antigen, and VPA dose or plasma concentration. VPA patients had significantly lower platelet counts and longer bleeding times than did controls (p < 0.05). Platelet count was inversely correlated to VPA dose and both free and total VPA concentration (p < 0.01). There was no significant difference in levels of von Willebrand's factor antigen between patients and controls. We assessed platelet aggregation by measurement of whole-blood platelet aggregation and release with agonists including adenosine diphosphate, thrombin, collagen, arachidonic acid, and ristocetin. VPA patients had significant decreases in platelet aggregation values compared with controls. There were significant differences in collagen, arachidonic acid, and adenosine diphosphate release and aggregation between groups that correlated to both VPA dose (p < 0.01) and concentration (p < 0.05). Prolongation of bleeding time was significantly correlated to both VPA dose and concentration. Our data suggest a significant relationship between impaired platelet function and both VPA dose and plasma concentration.