Background. The CAGE scale is a short test developed in the 1970s to screen for alcoholism or covert drinking problems. The reliability and validity of the scale has been demonstrated in the majority of studies conducted in clinical settings, but the validity of the scale in general population surveys has not yet been shown conclusively. Aims. The goal of this study was to assess the criterion validity of the CAGE scale in a general population survey. Method. Data from a large general health survey conducted in 1992 in the province of Quebec (Canada) (N=23 564) were used to define various measures of heavy drinking and drinking problems and to calculate, for male and female drinkers separately, standard measures of sensitivity, specificity and positive predictive value at different cutoff scores on the CAGE. Findings. With respect to both male and female drinkers, with all operational definitions of heavy drinking and useful cutoff scores on the CAGE, the scale was shown to be unable to discriminate between heavy drinkers and non-heavy drinkers. Prevalence of drinking problems among CAGE positive drinkers was also very low. Conclusion. These results do not support the use of the CAGE as a screening tool for heavy drinking and drinking problems in a general population survey or as a tool to estimate the prevalence of drinking problems in the population.