OBJECTIVES To compare, through cost-benefit analysis, a long-acting atypical antipsychotic (risperidone) with two second-generation oral antipsychotics (olanzapine and clozapine) and a “typical” prolonged action antipsychotic (pipotiazine). METHODS We designed a decision tree that included only direct costs, from the third party payer perspective and a single year time frame. Main epidemiological variables, obtained mostly from international literature and clinical trials, were adherence, relapses with and without hospitalization, and side effects. Local costs were obtained from Colombian health insurers and providers. RESULTS According to the model, the use of long acting risperidone injection was associated with the lowest overall costs to the system (US$5169 per year) followed closely by clozapine (US$5221) whose long term metabolic effects were not included in the model. Either one of these drugs would save around US$ 800 per year when compared with olanzapine (US$6087), and more than US$1000 when compared with pipotiazine (US$6544), the least expensive of the group and currently the standard of care in most patient groups. CONCLUSIONS Despite their apparent higher costs, atypical antipsychotics can be cost saving through their reduction in relapses and hospitalizations. The economic benefit, particularly with risperidone injection which had the lowest side effects, would be added to the direct and indirect benefits to the patients and their communities due mainly to relapse reduction.