Aims. To investigate the association between parental separation and alcohol consumption and problem drinking in early adulthood (at ages 23 and 33) . Design. The study used longitudinal data from the 1958 British birth cohort study, a large representative national sample followed to age 33 . Setting. Great Britain . Participants. 11 407 subjects were interviewed at age 33 in 1991 (69% of the target population). Analyses are based on 4606 men and 4892 women with data at ages 23 and 33 . Measurements. Units of alcohol consumed in the previous week at ages 23 and 33. Heavy drinking was defined as more than 20 units/week (women) and more than 35 units/week (men). Problem drinking was indicated by the four-item CAGE measure. Information on parental separation was reported by subjects at age 33; parental deaths were ascertained from data recorded in childhood sweeps of the survey . Findings. At age 23, the relationship between parental separation and alcohol consumption was weak and inconsistent, but by age 33 a stronger and more consistent relationship had emerged. Higher levels of alcohol consumption, heavy drinking and problem drinking (odds ratios 1.29-1.90) were found for those who had experienced parental divorce in childhood, but not later parental divorce or parental death. These results were not substantially attenuated by possible mediating factors, such as marital status or socio-economic circumstances . Conclusions. The risk associated with early parental divorce appeared to strengthen between ages 23 and 33. Life-course factors influencing the manifestation of alcohol problems in those from divorced families need to be identified.