Background: Previous studies of allopurinol and stroke risk have provided contradictory findings, ranging from a protective effect to an increased risk. Our objective was to assess whether allopurinol use is associated with the risk of stroke in the elderly. Methods: We used the 5 % random sample of Medicare beneficiaries from 2006-2012 to study the association of new allopurinol initiation and incident stroke. We used multivariable-adjusted Cox regression models adjusted for age, gender, race, Charlson index, and cardio-protective medications (beta-blockers, ACE inhibitors, diuretics, statins) to calculate hazards ratio (HR) with 95 % confidence intervals (CI). Sensitivity analyses adjusted for coronary artery disease (CAD) risk factors including hypertension, hyperlipidemia, diabetes, and smoking instead of Charlson index. Results: Among 28,488 eligible episodes of incident allopurinol, 2,177 ended in incident stroke (7.6 % episodes). In multivariable-adjusted analyses, allopurinol use was associated with 9 % lower hazard ratio for stoke, 0.91 (95 % CI, 0.83 to 0.99). Compared to no allopurinol use, allopurinol use durations of 181 days to 2 years, 0.88 (95 % CI, 0.78 to 0.99) and >2 years, 0.79 (95 % CI, 0.65 to 0.96) were significantly associated with lower multivariable-adjusted hazard of stroke. Sensitivity analyses adjusted for CAD risk factors confirmed these findings. In subgroup analyses, significant associations were noted between allopurinol use and the risk of ischemic stroke, 0.89 (95 % CI, 0.81 to 0.98); associations were not significant for hemorrhagic stroke, 1.01 (95 % CI, 0.79 to 1.29). Conclusions: Allopurinol use is associated with lower risk of stroke overall, more specifically ischemic stroke. This association is evident after 6-months of allopurinol use, and the hazard reduction increases with longer duration of use. Future studies need to examine underlying mechanisms.