The purpose of this investigation was to determine the effect of premenstrual edema (water retention) on measurements of percent body fat (%BF) in ovulating women during the menstrual cycle utilizing bioelectrical impedance. Specifically, this study was designed to investigate the difference between bioelectrical impedance measurements of %BF recorded during day 1 of menses, day 7, day -1 midcycle, midcycle, day 21, and days -3, -2, and -1 premenses. Determinations of impedance were made in 26 regularly menstruating women aged 20.23±0.74 yr using an electrical impedance analyzer with a four-electrode arrangement that induces a painless signal ( 800 !microamps at 50 kHz) into the body. Internal consistency reliability estimates made approximately 5 min apart during each test day ranged from R = .96-.99 for the impedance measurements of %BF, and the stability reliability correlation coefficients ever the eight observation days ranged from r = .92-.97 for the impedance measurements of %BF. The reliability analyses indicated that %BF esitimated from bioelectrical impedance was measured reliably. Specific days of the menstrual cycle did not significantly affect impedance measurements of %BF. Analysis of variance with repeated measures also indicated that day during the menstrual cycle did not significantly affect weight (Wt), %BF, resistance (Re), or total body water (TBW) measurements. However, urine osmolality (Osm) and basal body temperature (BBT) were significantly affected across days of the menstrual cycle (p < .01). A dramatic decline in Osm was noted at midcycle followed by a rise which peaked at day -3 premenses. After day -3 premenses, Osm rapidly declined at menses. The dramatic decline in Osm at midcycle may reflect changes in urine concentration due to actions of elevated estrogen associated with ovulation. BBT dropped prior to midcycle and then began to rise until day -3 of the next menses and then it dropped slightly. The drop prior -co midcycle may be also reflected by changes in the hormonal concentration of estrogen. The biphasic BBT response was considered to be presumptive evidence that ovulation had occurred. A 2 x 2 factorial analysis of variance with repeated measures demonstrated that there was a significant interaction between cycle length and activity level on measurements of Wt, %BF, and BBT (p < .05). As light activity subjects with average cycle lengths became moderately active, their %BF decreased from 26.51±0.63% to 21.11±0.50% followed by a dramatic increase to 29.40±0.62% as they became highly active. On the otherhand, as light activity subjects with above average cycle lengths became more and more active, their %BF dropped drastically from 35.11±1.44% to 18.72±0. 75%. The data support the hypothesis that bioelectrical impedance measurements of %BF recorded during the normal menstrual cycle are not significantly different. However, a larger population is required to validate the applicability of these results. / M.S.