Serum levels of total and free testosterone and 17 beta-estradiol were determined in 144 men with acute ischemic stroke and 47 healthy male control subjects. Blood samples from patients were drawn a mean of 3 days after stroke onset and also 6 months after admission in a subgroup of 45 patients. Initial stroke severity was assessed on the Scandinavian Stroke Scale and infarct size by computed tomographic scan. Mean total serum testosterone was 13.8 +/- 0.5 nmol/L in stroke patients and 16.5 +/- 0.7 nmol/L in control subjects (P = .002); the respective values for free serum testosterone were 40.8 +/- 1.3 and 51.0 +/- 2.2 pmol/L (P = .0001). Both total and free testosterone were significantly inversely associated with stroke severity and 6-month mortality, and total testosterone was significantly inversely associated with infarct size. The differences in total and free testosterone levels between patients and control subjects could not be explained by 10 putative risk factors for stroke, including age, blood pressure, diabetes, ischemic heart disease, smoking, and atrial fibrillation. Total and free testosterone levels tended to normalize 6 months after the stroke. There was no difference between patients and control subjects in serum 17 beta-estradiol levels. These results support the idea that testosterone affects the pathogenesis of ischemic stroke in men.