Determinations of total body water (TBW) calculated from deuterium dilution spaces and bioelectrical-impedance measurements were made serially in a group of 15 women before, during, and after pregnancy. Similar measurements were made once in a group of 50 nonpregnant women and intermittently in another group of 10 women during pregnancy and postpartum. TBW increased significantly during pregnancy, then decreased postpartum. Estimates of TBW in pregnancy and postpartum calculated with models derived from nonpregnant and pregnant women were similar to measured values. Changes in reactance and resistance explained more of the variance in predicting changes in TBW than did body weight, abdominal circumference, or hematocrit (50-75% vs 4-50%, respectively). Changes in TBW estimated with the nonpregnancy impedance model were significantly different than either the measured changes or changes predicted with the pregnancy impedance model. These findings indicate that the impedance method is a practical and valid method for determining longitudinal changes in TBW.