ObjectiveTo determine whether the increased acquisition costs associated with the atypical antipsychotic risperidone are offset by reductions in other mental health care utilisation and expenditure.Design and settingThe study was population-based and used South Carolina Medicaid claims data to determine changes in mental healthcare utilisation and expenditures related to schizophrenia. Changes in mental health−related utilisation and expenditures over time were calculated; total mental health−related expenditures and utilisation were disaggregated into pharmaceuticals, inpatient hospitalisations, and ambulatory and inpatient physician services [Health Care Financing Administration (HCFA) 1500 claims]. Groups of patients were compared for two 6-month periods preceding the initial prescription (pre1 and pre2), and two 6-month periods following the initial prescription (post1 and post2). Costs were discounted to the index date.PerspectivePayor (South Carolina Medicaid).PatientsThose patients with schizophrenia who received initial prescriptions for risperidone (n = 862), haloperidol (n = 325) or clozapine (n = 66) between February 1994 and June 1995 (index date).Main outcome measures and resultsThe mean increase in level of expenditure per person for pharmaceuticals from the pre- to the post-treatment period was significantly greater in the risperidone [751 US dollars ($US)] and clozapine ($US1423) groups than in the haloperidol group ($US6). However, the change in mean level of total mental healthcare expenditure per person was not significantly different for the risperidone group ($US832) compared with the haloperidol group ($US540) over the same time period, but the increase in the clozapine group was significantly higher ($US2500.23; p < 0.0001 for clozapinevsrisperidone and clozapinevshaloperidol).As the difference between the risperidone and haloperidol groups in pharmaceutical expenditures was not reflected in total mental healthcare expenditures, the remaining component costs were investigated to identify where the difference was offset. Compared with haloperidol, risperidone had a significantly smaller change in per person mean level of ambulatory and inpatient physician services claims for expenditure ($US692vs$US269, p = 0.01) and utilisation (+1.70vs−0.21, p < 0.0001).ConclusionsBased on these findings, we conclude that, in this population of patients with schizophrenia increased costs associated with risperidone were offset by decreases in other mental healthcare utilization. Risperidone is a technical substitute for ambulatory healthcare services.