Aims. To establish whether the high prevalence of alcohol abuse among unemployed people is explained by alcohol abuse causing unemployment, or vice versa. Design. A 5-year postal follow-up survey of a community sample of unemployed from Grenland, southern Norway. Participants. Two hundred and twenty-eight unemployed people, registered for more than 12 weeks, aged 16 to 63 years. Response rate 74%. Measurements. The Alcohol Use Disorder Identification Test (AUDIT) and DSM-III diagnoses of alcohol disorders in medical examinations. Findings. At the 5-year follow up, 23% of those still unemployed and 12% of those re-employed scored higher than the AUDIT cut-point of 10. Re-employment reduced the chance of scoring positive on the AUDIT to 34% of the chance for those still unemployed. Significant selection to long-term unemployment according to AUDIT score was not demonstrated. None of the 7% who had a DSM-III diagnosis of an alcohol disorder had a job 5 years later, however, suggesting that alcohol-related selection to unemployment does occur. Conclusion. The high prevalence of harmful drinking among Norwegian unemployed is explained mainly by unemployment causing alcohol abuse rather than vice versa. Reducing unemployment should contribute to reduced alcohol problems in Norway.