OBJECTIVE To investigate the relationship between single-frequency bioimpedance at 50 kHz (both total body impedance and segmental impedance) and total body water, measured using tritiated water in the presence of the severe fluid retention seen in multiple organ failure. METHODS Prospective, controlled study. METHODS General intensive care unit. METHODS Twenty patients with multiple organ failure and 30 normal volunteers, of whom a subgroup of ten had total body water measured. METHODS None. RESULTS Total body water and total and segmental bioimpedance values at 50 kHz were measured using tritiated water and a Holtain body composition analyzer in 20 patients with multiple organ failure and in ten normal volunteers. An additional 20 normal volunteers also had segmental and total body impedance measured. There was no difference in the linear regression lines constructed for the patients and the volunteers, but the SEM in the patients (7.6 L) was four times that seen in the normal subjects (1.9 L). In a further seven patients, the impedance technique overestimated the change in total body water, deduced from acute changes in weight, by between 0% and 46% (median, 12%). In the patients, who were supine, the knee-ankle segment contributed significantly more to total body impedance than it did in the normal volunteers (42.5% vs. 34.4%; p < .001), who were upright and mobile immediately before the measurement. CONCLUSIONS Although single-frequency bioimpedance does give an indication of total body water and change in total body water, it is neither precise nor accurate enough to be the sole guide to fluid therapy. The proportion of total impedance contributed by the knee-ankle segment, which contains relatively little water, was significantly greater in the patients than in the controls, probably reflecting better drainage of fluid from the lower limb in the supine position.