Because body fat distribution has been recognized as a disease risk factor, practical methods for the measurement of body girths are needed. In two groups of postmenopausal women aged 55-69 years in the upper midwestern United States, the authors examined the reliability and accuracy of self-measurement by mail questionnaire of waist, hip, upper arm, wrist and calf girths. Intra-class correlations for waist girth were 0.96 when two self-measurements were compared and 0.93 when self-measurement was compared with technician measurement. Other intra-class correlations were at least 0.85 for repeat self-measurements except for wrist, which had an intra-class correlation of 0.66. For comparisons of self-measurement with technician measurement, intra-class correlations ranged from 0.71 for upper arm to 0.96 for hips. There was slight overestimation of waist girths and underestimation of hip girths when self-measurement was compared with technician measurement. Accuracy of self-measurement did not seem to vary according to age or educational status, but for hip, wrist, and calf girths it appeared that self-measurement underestimated technician measurement as girth size increased. For most girths, within-person variation in girth measurement also increased as girth size increased. Overall, girth self-measurement was both repeatable when re-ascertained by mail and accurate when compared with subsequent technician measurement.