Background and purpose: Numerous case series have proposed a relationship between 'crack' cocaine use and stroke. We performed a retrospective case control study at a large inner-city public hospital to determine the relationship between crack use and stroke among young persons. Methods: We reviewed records of all patients aged 20 to 39 years with a diagnosis of stroke, and of controls selected from patients with noncocaine- related diagnoses, admitted from January 1990 through June 1994. We collected information regarding cocaine use, time of last use, route of use, and the results of urine toxicologic studies. We performed backward stepwise logistic regression analyses to determine the association of crack use at any time and acute crack use (defined as use within 48 hours prior to presentation) with stroke and stroke subtypes. Results: Among patients with information regarding presence or absence of crack use (66 of 144 stroke patients and 99 of 147 controls), crack use at any time was not associated with stroke (odds ratio [OR] = 0.7, 95% CI 0.4-1.8) or cerebral infarction (OR = 0.5, 95% CI 0.2-1.2). Among patients providing temporal information regarding crack use, acute crack use was not associated with stroke (OR = 1.9, 95% CI 0.7-5.1) or cerebral infarction (OR = 1.2, 95% CI 0.4-3.8). Conclusions: Crack use at any time or acute crack use was not significantly associated with stroke or cerebral infarction in our patient population.