The purported advantages of second-generation or "atypical" antipsychotics relative to first-generation antipsychotics have not been examined in patients with a first episode of schizophrenia. This flexible-dose study examined efficacy and safety in a randomized, double-blind, 52-week trial, comparing chlorpromazine (CPZ) and clozapine (CLZ) in treatment naive patients experiencing their first episode of schizophrenia. In all, 160 inpatients with first-episode schizophrenia or schizophreniform disorder were randomized to CPZ or CLZ and followed them for 52 weeks or until dropout. The primary efficacy measure was time to first remission and proportion of time remaining in remission. The analysis was supplemented by comparisons on a profile of clinical symptoms and side effects. Of these first-episode patients, 80% achieved remission within 1 year (79% CPZ, 81% CLZ). The Kaplan-Meier estimated median time to first remission was 8 weeks for CLZ vs 12 weeks for CPZ (chi(2)(1)=5.56, p=0.02). Both the rate of first achieving remission and the odds for being in remission during the trial were almost doubled for the CLZ group in comparison with the CPZ group. At 12 weeks, CLZ was superior on many rating scale measures of symptom severity while CPZ was not superior on any. These symptom differences remained significant when controlling for EPS differences. By 52 weeks many of the symptom differences between groups were no longer significantly different. Generally, CLZ produced fewer side effects than CPZ, particularly extrapyramidal side effects. There was no significant difference between treatments in weight change or glucose metabolism. For each prior year of untreated psychosis, there was a 15% decrease in the odds of achieving remission (OR=0.85; CI 0.75-0.95). A high proportion of first-episode patients remitted within 1 year. We detected no difference in the proportion of first-episode patients receiving CLZ or CPZ that achieved remission. However, first-episode patients receiving CLZ remitted significantly faster and remained in remission longer than subjects receiving CPZ. While the CLZ group showed significantly less symptomatology on some measures and fewer side effects at 12 weeks, the two treatment groups seemed to converge by 1 year. Longer duration of untreated psychosis was associated with lower odds of achieving remission.