BACKGROUND: Psychological stress and alcohol are both suggested as risk factors for stroke. Further, there appears to be a close relation between stress and alcohol consumption. Several experimental studies have found alcohol consumption to reduce the immediate effects of stress in a laboratory setting. We aimed to examine whether the association between alcohol and stroke depends on level of self-reported stress in a large prospective cohort. METHODS: The 5,373 men and 6,723 women participating in the second examination of the Copenhagen City Heart Study in 1981-1983 were asked at baseline about their self-reported level of stress and their weekly alcohol consumption. The participants were followed-up until 31st of December 1997 during which 880 first ever stroke events occurred. Data were analysed by means of Cox regression modelling. RESULTS: At a high stress level, weekly total consumption of 1-14 units of alcohol compared with no consumption seemed associated with a lower risk of stroke (adjusted RR: 0.57, 95% CI: 0.31-1.07). At lower stress levels, no clear associations were observed. Regarding subtypes, self-reported stress appeared only to modify the association between alcohol intake and ischaemic stroke events. Regarding specific types of alcoholic beverages, self-reported stress only modified the associations for intake of beer and wine. CONCLUSIONS: This study indicates that the apparent lower risk of stroke associated with moderate alcohol consumption is confined to a group of highly stressed persons. It is suggested that alcohol consumption may play a role in reducing the risk of stroke by modifying the physiological or psychological stress response.