The hypothesis that a single measurement, waist circumference, might be useful to identify people at health risk both from being overweight and from having central obesity was tested using data on 1513 subjects from a prevalence survey of diabetes mellitus in Hong Kong Chinese. It was found that a waist circumference > or = 94 cm for men and > or = 80 cm for women identified subjects with high BMI (> or = 25 kg/m2), and those with lower BMI but high WHR (> or = 0.95 for men, > or = 0.80 for women) with a sensitivity of < 31% and specificity of 100%. Decreasing the waist circumference cut-off to 85 cm for men and 75 cm for women increased the sensitivity to 79.2% in men and 56.4% in women. It was concluded that a single waist circumference measurement did not allow sensitive identification of people at health risk from being overweight or from having central obesity.