OBJECTIVES: This study compares switching and discontinuation rates for patients who had already switched from fluoxetine to other branded selective serotonin reuptake inhibitors (SSRIs). METHODS: Retrospective analysis of claims in an administrative database from January 1, 1999 to June 30, 2002. Claims for patients on fluoxetine who switched to sertraline, paroxetine or citalopram were analyzed to compare rates of discontinuation or second switch to another SSRI. We used Cox proportional models and life table survival curves. Definition of switching or discontinuation was based on SSRI refill within 15 days or 2 days supply. RESULTS: Paroxetine patients (N = 217) were significantly more likely to switch (HR = 1.47, p = 0.033) than sertraline patients (N = 227). The likelihood of switching (HR = 0.98, p = 0.917) was not significantly different for patients on citalopram (N = 291) than on sertraline. Patients were as likely to discontinue when on sertraline (N = 324), paroxetine (N = 293) or citalopram (N = 437). Average number of distinct drugs (p = 0.0005) and average number of claims (p = 0.0006) were significant predictor of discontinuation, while age, gender and co-payment were not. CONCLUSIONS: Patients on fluoxetine who had switched to paroxetine were significantly more likely to switch again than those who had switched to sertraline. The likelihood of discontinuation was similar among patients on sertraline, paroxetine or citalopram.