BACKGROUND: Substantial weight gain is common with many atypical antipsychotics. AIMS: To evaluate the extent, time course and predictors of weight gain and its effect on study retention among people with first-episode psychosis treated with olanzapine or haloperidol. METHOD: Survival analysis assessed time to potentially clinically significant weight gain (> or =7%) and the effect of weight gain on study retention. Weight gain during the 2-year study was summarised using last-observation-carried-forward (LOCF), observed cases and study completion approaches. RESULTS: After 2 years of treatment, LOCF mean weight gain was 10.2 kg (s.d.=10.1) for olanzapine (n=131) and 4.0 kg (s.d.=7.3) for haloperidol (n=132); observed cases mean weight gain was 15.4 kg (s.d.=10.0) for olanzapine and 7.5 kg (s.d.=9.2) for haloperidol. Change in body mass index was significantly predicted only by treatment group (P < 0.0001). CONCLUSIONS: Olanzapine was associated with significantly greater weight gain than haloperidol, with both leading to greater weight gain than previously described.