The medical record linkage system for the Rochester Epidemiology Project provided the means to identify 1,444 incidence cases of ischemic stroke and age- and sex-matched controls from the population from 1960 to 1984 to conduct a case-control study nested in the population. A multiple logistic-regression model permitted the estimation of odds ratios of ischemic stroke for each risk factor while adjusting for confounding variables. The final model, in addition to age and date of stroke, included transient ischemic attacks, hypertension, current smoking, atrial fibrillation, ischemic heart disease, mitral valve disease (other than prolapse), and diabetes mellitus. The process identified interactions showing that ischemic stroke incidence for persons with transient ischemic attacks was higher in women than in men and that the risk decreased with increasing age; that the risk of stroke with hypertension and also with current cigarette smoking decreased with increasing age; and that the risk of ischemic stroke with intermittent or persistent atrial fibrillation was similar when hypertension was present, but without hypertension the risk of stroke was more than seven times greater with persistent than with intermittent atrial fibrillation. None of the odds ratios differed over the five quinquennia of the study, and no effect of antihypertensive treatment on stroke incidence could be demonstrated in the population.