To assess from a health sector perspective the incremental cost-effectiveness of eight drug treatment scenarios for established schizophrenia. Using a standardized methodology, costs and outcomes are modelled over the lifetime of prevalent cases of schizophrenia in Australia in 2000. A two-stage approach to assessment of health benefit is used. The first stage involves a quantitative analysis based on disability-adjusted life years (DALYs) averted, using best available evidence. The robustness of results is tested using probabilistic uncertainty analysis. The second stage involves application of 'second filter' criteria (equity, strength of evidence, feasibility and acceptability) to allow broader concepts of benefit to be considered. Replacing oral typicals with risperidone or olanzapine has an incremental cost-effectiveness ratio (ICER) of 48,000 Australian dollars and 92,000 Australian dollars/DALY respectively. Switching from low-dose typicals to risperidone has an ICER of 80,000 Australian dollars. Giving risperidone to people experiencing side-effects on typicals is more cost-effective at 20,000 Australian dollars. Giving clozapine to people taking typicals, with the worst course of the disorder and either little or clear deterioration, is cost-effective at 42,000 Australian dollars or 23,000 Australian dollars/DALY respectively. The least cost-effective intervention is to replace risperidone with olanzapine at 160,000 Australian dollars/DALY. Based on an 50,000 Australian dollars/DALY threshold, low-dose typical neuroleptics are indicated as the treatment of choice for established schizophrenia, with risperidone being reserved for those experiencing moderate to severe side-effects on typicals. The more expensive olanzapine should only be prescribed when risperidone is not clinically indicated. The high cost of risperidone and olanzapine relative to modest health gains underlie this conclusion. Earlier introduction of clozapine however, would be cost-effective. This work is limited by weaknesses in trials (lack of long-term efficacy data, quality of life and consumer satisfaction evidence) and the translation of effect size into a DALY change. Some stakeholders, including SANE Australia, argue the modest health gains reported in the literature do not adequately reflect perceptions by patients, clinicians and carers, of improved quality of life with these atypicals.