There are indications that body fat distribution, independent of obesity, may be a disease risk factor. The accuracy of self-measurement of waist, hip, and thigh circumferences, indicators of fat distribution, were examined for 227 women (aged 19-45) who were sent either a calibrated or uncalibrated tape measure and subsequently remeasured by a trained technician. Subjects tended to underestimate all circumferences somewhat. Spearman correlations between self-measurement and technician measurement were 0.88, 0.89, and 0.86 for the waist, hip, and thigh, respectively, and 0.65 and 0.77 for the waist-to-hip waist-to-thigh ratios. Correlations for waist, but not other measures, were better with the uncalibrated than with the calibrated tape. Accuracy did not vary within strata of education, body mass index, or waist-to-hip ratio. Subjects were able to measure their circumferences with reasonable accuracy. Circumference measures were more accurate than were computed ratios of circumferences; waist-to-thigh ratios were more accurate than were waist-to-hip ratios.