A study was designed to define the osmol gap in patients whose serum ethanol concentrations are known, to reevaluate several accepted equations for calculating osmolarity, and to apply the results to the theoretical clinical scenario of a toxic alcohol ingestion. The design for the study used consecutive, prospective enrollment of all patients presenting to a large inner city hospital who clinically required determination of their serum ethanol and electrolytes. Three hundred and twenty one consecutive adult patients were enrolled in the study, sixteen were excluded from the final analysis. A stepwise multiple linear regression analysis was performed to determine the best coefficients for sodium, blood urea nitrogen, and ethanol from the data set. Osmolarity was then calculated using these coefficients and traditional models. The osmol gap (measured osmolality minus calculated osmolarity [2*Na + BUN/2.8 + Glu/18 + Etoh/4.6]) was -2 +/- 6 mOsm. Although different equations produced different osmol gaps (ranging from -5 to + 15 mOsm) the standard deviations and correlation coefficients were similar. Large variations exist in the range of osmol gaps. Absolute values are very dependent on the equations used to calculate osmolarity. Because of the larger range of values, small osmol gaps should not be used to eliminate the possibility of toxic alcohol ingestion.