A decision-tree simulation model is used to examine the costs associated with olanzapine versus haloperidol in the treatment of patients with schizophrenia in the UK. Parameter values and outcome scores were derived mainly from an international clinical trial. Resource consequences were examined on the basis of assumed service delivery and actual unit costs specific to the UK. While olanzapine is more expensive to prescribe than haloperidol, it generates savings by reducing utilisation of medical services. As a result, a comparison of the 2 drugs is approximately cost neutral. Model uncertainties are examined using extensive sensitivity analysis; in most scenarios, cost-neutral results are maintained. Olanzapine is more effective than haloperidol as measured by Brief Psychiatric Rating Scale scores and non-relapse rates. With such gains in effectiveness and near equivalence in terms of costs, olanzapine, in comparison with haloperidol, may represent a cost-effective treatment option.