OBJECTIVE To assess the value of bioimpedance as a clinical tool by determining the accuracy and bias of single and multiple frequency bioimpedance estimates of total body and extracellular water in comparison with values established by criterion reference techniques. METHODS Controlled, prospective, single-blind investigation. METHODS Private, not-for-profit, university-affiliated, acute care hospital. METHODS Eight male, post-elective coronary artery bypass graft surgical patients. METHODS Within 6 hrs after surgery, estimates of total body and extracellular water volumes were determined using single and multiple frequency bioimpedance techniques. These estimates were then compared with the gold standard volumes measured by deuterium oxide and bromine dilutional space determination, respectively. RESULTS The mean multiple frequency bioimpedance estimate of total body water of 47.7 +/- 9.4 L was statistically different from the single frequency bioimpedance and deuterium values of 52.5 +/- 9.4 (p < .006) and 53.3 +/- 11.6 L (p < .002), respectively. In comparison, the mean multiple and single frequency bioimpedance estimates of extracellular water, 26.3 +/- 5.4 and 29.2 +/- 5.4 L, respectively, were not statistically different from the bromine value of 27.5 +/- 6.9 L. In addition, the mean errors for multiple and single frequency bioimpedance determinations of extracellular water, -1.2 +/- 2.0 and 1.7 +/- 2.7 L, respectively, were statistically different (p = .001). CONCLUSIONS In male, post-elective coronary artery bypass graft surgical patients, single frequency bioimpedance was a more accurate and less biased predictor of total body water than multiple frequency bioimpedance. The accuracy and bias of multiple frequency bioimpedance was superior to single frequency bioimpedance for the prediction of extracellular water. Whether this observation remains true for other populations of critically ill patients remains to be investigated.