It has been suggested in the literature that cut-off points based on waist circumference (waist action levels) should replace cut-off points based on body mass index (BMI) and waist-to-hip ratio in identifying subjects with overweight or obesity. In this article, we examine the sensitivity and specificity of the cut-off points when applied to 19 populations with widely different prevalences of overweight. Our design was a cross-sectional study based on random population samples. A total of 32,978 subjects aged 25-64 years from 19 male and 18 female populations participating in the second MONICA survey from 1987 to 1992 were included in this study. We found that at waist action level 1 (waist circumference > or =94 cm in men and > or =80 cm in women), sensitivity varied between 40% and 80% in men and between 51% and 86% in women between populations when compared with the cut-off points based on BMI (> or =25 kg/m2) and waist-to-hip ratio (> or =0.95 for men, > or =0.80 for women). Specificity was high (> or =90%) in all populations. At waist action level 2 (waist circumference > or =102 cm and > or =88 cm in men and women, respectively, BMI > or =30 kg/m2), sensitivity varied from 22% to 64% in men and from 26% to 67% in women, whereas specificity was >95% in all populations. Sensitivity was in general lowest in populations in which overweight was relatively uncommon, whereas it was highest in populations with relatively high prevalence of overweight. We propose that cut-off points based on waist circumference as a replacement for cut-off points based on BMI and waist-to-hip ratio should be viewed with caution. Based on the proposed waist action levels, very few people would unnecessarily be advised to have weight management, but a varying proportion of those who would need it might be missed. The optimal screening cut-off points for waist circumference may be population specific.