The relation between obesity, particularly abdominal obesity, and risk of stroke amongst women remains unclear. In 1991-1992, a prospective study was initiated in Sweden amongst women who returned a self-administered questionnaire. Through linkage with nation-wide registries, 45,449 women, free of stroke at entry, were followed up until diagnosis of first incident stroke, death, or the end of follow-up in 2002. We estimated multivariate relative risks (RRs) with 95% confidence intervals (CIs) from Cox proportional hazards regression models. A total of 170 incident stroke cases occurred during an average of 11 years of follow-up. The RR of stroke amongst women in the highest compared with the lowest quintile was 2.4 (95% CI 1.3-4.2; P for trend 0.04) for waist-to-hip ratio, 2.5 (95% CI 1.5-4.3; P for trend 0.01) for waist-to-height ratio and 2.3 (95% CI 1.2-4.3; P for trend 0.02) for waist circumference. Adjustment for hypertension and diabetes attenuated these risk estimates. In contrast, birth weight, body mass index (BMI) at age 18, BMI at entry, weight change in adulthood and adult height were not significantly associated with risk of stroke. This study provides evidence that, in contrast to BMI, several different measures of abdominal obesity are strong predictors of stroke in women.