Data on the long-term metabolic side-effects associated with antipsychotics are scarce. Prospective longitudinal studies in medication-naive patients with a first episode of psychosis are a valuable source of information as they provide an assessment prior to the antipsychotic exposure and minimize the effect of potential confounding factors. The aim of this study was to assess the course of weight gain and the incidence of metabolic abnormalities during the first 3 yr of antipsychotic treatment. Data were collected from a cohort of 170 first-episode psychosis patients. They were randomly assigned to haloperidol (32%); olanzapine (32%) and risperidone (36%). The dose used was flexible. The initial antipsychotic treatment was changed when required, based on clinical response and tolerability. The results showed that the mean weight gain at 3 yr was 12.1 kg (s.d. = 10.7). It appeared to increase rapidly during the first year (85% of the total mean weight gain) and then stabilized gradually over time. Total cholesterol, LDL-cholesterol and triglyceride levels followed a similar trajectory with a significant increase only during the first year. No significant changes were detected in the mean values of glycaemic parameters. Two patients with a family history of diabetes developed diabetes type II. At short-term the factors positively associated with weight gain were lower body mass index, male gender and olanzapine treatment. At long-term, functional status and clinical response were the main predictors. The results of our study indicate that the first year of antipsychotic treatment is a critical period for weight gain and metabolic changes. Identification of weight gain patterns may help to inform studies that aim to prevent or mitigate the metabolic adverse events associated with antipsychotic therapy.