Previous studies suggest that problem drinkers have markedly increased sickness absence. However, it is not clear how more moderate alcohol consumption and abstinence relate to sickness absence. As part of the Whitehall II study the relationship between different drinking patterns and sickness absence is examined. A total of 10,314 male and female civil servants completed a baseline questionnaire about their drinking habits--the amount of alcohol consumed over the past 7 days and the frequency of drinking over the past 12 months. All sickness absence has been recorded prospectively. Drinking patterns and sickness absence are examined for short spells (< or = 7 days) and long spells (> 7 days) adjusting for other causes of sickness absence: age, grade of employment, smoking, work characteristics and baseline health. Alcohol consumption was strongly related to employment grade, the lower the grade the higher proportion of men and women reporting no alcohol consumption. For men the relation of alcohol intake to short spells of sickness absence (< or = 7 days) appeared to be U-shaped, for long spells (> 7 days) increased rates of absence were found only in frequent drinkers. There was no clear relationship for women, however higher rates of sickness absence were found in non-drinkers.