The present study was designed to compare direct health service expenditures, for the treatment of depression, among patients enrolled in a health maintenance organization, and prescribed one of three selective serotonin reuptake inhibitors, fluoxetine, paroxetine or sertraline. Information regarding depression-related health service use was derived from the computer archive of a network-model health maintenance organization system serving 700,000 beneficiaries. A total of 744 health maintenance organization beneficiaries were found to satisfy the study selection criteria. Multivariate regression analysis was used to determine the incremental influence of selected demographic, clinical, financial and provider characteristics on health service expenditures related to the treatment of depression (ICD-9-CM, or DSM-IV code 296.2) 1 year after the start of antidepressant pharmacotherapy. Multivariate findings indicate that treatment with paroxetine increases average expenditures for physician visits ($31.93; P < or = 0.05), psychiatric visits ($19.33; NS), laboratory tests ($2.35; P < or = 0.05), hospitalizations ($85.33; P < or = 0.05), psychiatric hospitalizations ($82.01; P < or = 0.05), and antidepressant pharmacotherapy ($63.72; P < or = 0.05), for a total per capita increase in health service use of $284.68 (P < or = 0.05), compared with treatment with fluoxetine. Sertraline treatment increases average expenditures for physician visits ($21.74; P < or = 0.05), psychiatric visits ($56.79; P < or = 0.05), laboratory tests ($1.21; P < or = 0.05), hospitalizations ($70.59; P < or = 0.05), psychiatric hospitalizations ($95.75; P < or = 0.05), and antidepressant pharmacotherapy ($69.85; P < or = 0.05), for a total per capita increase in health service use of $315.96 (P < or = 0.05), compared with treatment with fluoxetine. Economic comparisons between paroxetine and sertraline did not demonstrate any significant differences in expenditures for the health services examined.