We used the Kaplan-Meier product limit method to estimate rates and Cox proportional hazards regression analysis with bootstrap validation to model significant independent predictors of and temporal trends in survival and recurrent stroke among 1,111 residents of Rochester, MN, who had a first cerebral infarction from 1975 through 1989. The risk of death after first cerebral infarction was 7% +/- 0.7% at 7 days, 14% +/- 1.0% at 30 days, 27% +/- 1.3% at 1 year, and 53% +/- 1.5% at 5 years. Independent risk factors for death after first cerebral infarction were age (p < 0.0001), congestive heart failure (p < 0.0001), persistent atrial fibrillation (p < 0.0001), recurrent stroke (p < 0.0001), and ischemic heart disease (p < 0.0001 for age < or =70, p > 0.05 for age >70). The risk of recurrent stroke after first cerebral infarction was 2% +/- 0.4% at 7 days, 4% +/- 0.6% at 30 days, 12% +/- 1.1% at 1 year, and 29% +/- 1.7% at 5 years. Age (p = 0.0002) and diabetes mellitus (p = 0.0004) were the only significant independent predictors of recurrent stroke. Neither the year nor the quinquennium of the first cerebral infarction was a significant determinant of survival or recurrence. The temporal trend toward improving survival after first cerebral infarction documented in Rochester, MN, in the decades before 1975 has ended.