Aim. To assess the possible effects of changes in marital status, employment status and having children at home on alcohol consumption and the frequency of heavy drinking. With role theory as a starting point it was expected that a shift into more social roles would decrease consumption and heavy drinking while the shift away from social roles would be associated with an increase in consumption and heavy drinking. Design. Prospective cohort study. Setting. The province of Limburg, The Netherlands (1980-89). Participants. 1327 men and women aged 16-69 years at first measurement. Measurements. Weekly consumption of standard units (10 g ethanol) of alcoholic beverages; frequency of drinking six units or more; self-reported social role. Findings. The acquisition of a spouse role and a parental role but not an employment role was associated with a decrease in consumption or heavy drinking. The loss of the spouse role among women was associated with an increase in heavy drinking. Otherwise, losing a role was not linked with a change in consumption and heavy drinking. Conclusions. Limited support was found for the expectation that role transitions influence drinking behaviour. Our study suggests that other theories must be sought to explain social differences in drinking behaviour.