In earlier studies, we concluded that biliary cholesterol supersaturation may be a necessary but not sufficient cause for gallstone formation. In the present studies, we calculated cholesterol saturation indices (CSI) for 120 bile specimens of cholesterol gallstone patients and controls, using the solubility boundaries of Carey and Small (J Clin Invest 61:998-1026, 1978) for artificial biles. The mean CSI + SD was 1.42 +/- 0.68 for controls and 1.80 +/- 1.02 for cholesterol gallstone patients. Of the control bile samples, 68% were supersaturated. Since the two groups could not be sharply distinguished by saturation index, we studied another property of bile samples: nucleation time for cholesterol crystal formation. The mean nucleation time for 16 control bile samples was 15 days, and a strong correlation (r = 0.84) was found between CSI and nucleation time. Twelve bile samples of cholesterol gallstone patients had a mean nucleation time of 3 days and no correlation between CSI and nucleation time. A discriminant analysis of the data, which took into account both the CSI and the nucleation time for each sample, allowed a sharp distinction between bile samples of cholesterol gallstone patients and controls. This distinction could not have been drawn from studies with artificial biles, since these differ only in their CSI.