OBJECTIVES: This study assesses predictors of stroke recurrence in a Medicaid population, specifically examining persistence on clopidogrel, warfarin and aspirin drug therapy. METHODS: Medical and pharmacy claims for patients with a diagnosis of stroke (ICD-9 Codes 430-438) were obtained from Medicaid for the period of 1/1/01-12/31/03. Only patients with at least one month of follow up were included. In order to obtain an incident cohort, patients who had a stroke diagnosis between 1/1/01-6/30/01 were excluded. Cox proportional hazard models were built to predict the time dependent likelihood of no recurrence, as a function of drug persistence, for patients who start on clopidogrel, warfarin or aspirin, adjusting for blood pressure lowering and lipid lowering drugs taken after the first stroke, hypertension, heart disease, diabetes, race, gender, age and residence (urban or not). RESULTS: In total, there were 925 patients, 36% older than 60, 58% African Americans, and 35% males. Patients who persist on the first drug taken post-stroke (HR=1.29, p=0.03, CI 1.02-1.63), start on clopidogrel (HR=1.42, p=0.006 CI 1.10-1.82) or take up to two (HR=1.43, p=0.001, CI 1.15-1.78) blood pressure lowering drugs post-stroke are significantly more likely to have no recurrence. Patients with a diagnosis of heart disease (HR=0.81, p=0.05, CI 0.66-0.99), hypertension (HR=0.64, p=0.00, CI 0.51-0.79) or diabetes (HR=0.75, p=0.006, CI 0.61-0.92) post-stroke are less likely to have no recurrence, i.e. to stay in remission after the first stroke. CONCLUSIONS: Among Medicaid patients, persistence on the first drug taken after a stroke reduces the likelihood of recurrence. Clopidogrel and the combination of up to two blood pressure lowering drugs are associated with a higher likelihood of no recurrence. However, heart disease, hypertension and diabetes decrease the likelihood of avoiding a second stroke. These results may inform care management plans for Medicaid patients.