The association of cocaine and amphetamine use with hemorrhagic and ischemic stroke is based almost solely on data from case series. The limited number of epidemiologic studies of stroke and use of cocaine and/or amphetamine have been done in settings that serve mostly the poor and/or minorities. This case-control study was conducted in the defined population comprising members of Kaiser Permanente of Northern and Southern California. We attempted to identify all incident strokes in women ages 15-44 years during a 3-year period using hospital admission and discharge records, emergency department logs, and payment requests for out-of-plan hospitalizations. We selected controls, matched on age and facility of usual care, at random from healthy members of the health plan. We obtained information in face-to-face interviews. There were 347 confirmed stroke cases and 1,021 controls. The univariate matched odds ratio for stroke in women who admitted to using cocaine and/or amphetamine was 8.5 (95% confidence interval = 3.6-20.0). After further adjustment for potential confounders, the odds ratio in women who reported using cocaine and/or amphetamine was 7.0 (95% confidence interval = 2.8-17.9). The use of cocaine and/or amphetamine is a strong risk factor for stroke in this socioeconomically heterogeneous, insured urban population.