A growing body of literature suggests that the anticonvulsant valproate may be effective in some psychiatric disorders. We evaluated 73 consecutive psychiatric patients without diagnosable neurologic disorder, and refractory to previous medications, who were treated with valproate. The drug appeared: 1) moderately to markedly effective in 59% of patients in the manic phase of bipolar or schizoaffective disorder; 2) largely ineffective in patients with major depression, schizophrenia, and other psychiatric disorders; and 3) well tolerated, with no cases of serious hepatic or hematologic toxicity. Age, sex, duration of illness, presence of psychotic symptoms, and presence of soft neurological signs or CT scan abnormalities did not predict response to valproate; however, nonspecific (nonparoxysmal) EEG abnormalities showed a slight but significant association with response.
