Seventy-two newly readmitted, drug-free men with the diagnosis of schizophrenia by DSM-III were assigned randomly to receive fluphenazine hydrochloride at 5 mg, 10 mg, or 20 mg daily for 4 weeks. Fluphenazine (FLU), fluphenazine sulfoxide, 7-hydroxyfluphenazine, and fluphenazine N-oxide were measured by highly specific and sensitive radioimmunoassays. Data were analyzed by logistic regression using the Clinical Global Impressions Disabling Side Effects and Global Improvement as the outcome measures. Disabling side effects were defined as "side effects that significantly interfered with patient's functioning" or "side effects that outweigh therapeutic effects" (National Institute of Mental Health 1985, p. 839). Higher plasma FLU levels (up to 4.23 ng/mL) were significantly (p = .015) associated with a higher rate of global improvement. However, close to 90 percent of these acute patients had disabling side effects at a plasma FLU level of 2.7 ng/mL. At least in the patient's view, these disabling side effects negated or compromised the improvement in psychosis. Fluphenazine N-oxide may be a toxic metabolite in that it was more powerfully associated with side effects than was the parent FLU.