Predictors of the costs of psychosis have received insufficient research attention, particularly factors associated with indirect costs.

To identify the predictors of direct mental health care costs and indirect or time-loss costs in psychotic disorders and to discuss their implications for future interventions.

Structured interview data from the Low Prevalence Disorders Study (n=980) were used to examine predictors of the costs of psychosis in Australia. Estimates of annual costs per patient were derived from the perspectives of government and society. Hierarchical regressions were used to assess the contributions to costs of premorbid, psychosocial and clinical factors.

Schizophrenia involved greater costs than other psychotic disorders. Non-completion of high-school education and chronicity of illness course were predictive of higher costs across all categories, and some factors were linked primarily with mental health care costs (e.g. age at onset, current symptomatology) or indirect costs (e.g. male gender, overall disability).

Several concurrent strategies were recommended, including early intervention programmes and assertive evidence-based rehabilitation and supported employment programmes aimed at reducing disability The cost-effectiveness of these approaches needs to be evaluated from the perspectives of both government and society.