A representative sample of 310 long-term unemployed in Norway was followed for 2 years with clinical examinations and the AUDIT questionnaire. 30% of the men and 8% of the women scored over the cut-off point for an alcohol use disorder. This gives a probable prevalence of 16%. The test predicted return to employment in this sample. The AUDIT answers were also used as a basis for dividing into three groups: 'normal', 'hazardous' and 'harmful'. At 2 year follow-up, 27% had changed group, 32 respondents to the worse and 24 to the better. This 'unstable' group was characterized by weaker social network and more frequent drinking. The AUDIT was judged as a useful instrument both in a routine health examination and as an epidemiological tool.