Acute behavioral response to 20 mg of dextroamphetamine (intravenous) predicted fourth-week antipsychotic response to double-blind pimozide treatment. Patients whose psychotic condition improved with dextroamphetamine administration showed more antipsychotic response to pimozide therapy than those whose condition worsened or did not change. Multiple regression analysis indicated amphetamine-induced response predicted pimozide response after four weeks for fifth-week pimozide response was more accurately predicted by prepimozide psychosis ratings. Our study provides some evidence that mechanisms underlying early and late pimozide response are not necessarily identical. Because patients who did not respond to dextroamphetamine administration still improved with pimozide therapy, our data do not support the concept that schizophrenia can be divided into two groups (dopamine-sensitive or dopamine-insensitive) but, rather, that dopamine responsiveness changes over time. Clinical application is not warranted until studies with larger samples have replicated our findings.