OBJECTIVE To quantitate the contribution of lactate, phosphate, urate, total serum proteins, and unidentified anions to the anion gap in patients with severe sepsis. METHODS Thirty critically ill patients with evidence of severe sepsis and systemic hypoperfusion were prospectively studied. METHODS The anion gap was calculated as [Na+] + [K+] - [Cl-] - [HCO3]. A corrected anion gap was calculated as the anion gap minus the anionic contribution of lactate, phosphate, urate, and total serum proteins. The corrected anion gap is a marker of unmeasured anion less unmeasured cation concentration. RESULTS The mean anion gap was 21.8 +/- 1.4 mmol/L and the corrected anion gap was 3.7 +/- 0.8 mmol/L. The mean arterial blood lactate concentration was 5.9 +/- 0.8 mmol/L. The magnitude of the lactate concentration correlated linearly with the anion gap (r2 = .61, lactate = 0.4 anion gap - 3.9, n = 30, p less than .01). The corrected anion gap was greater than 0 in 24 (80%) of 30 patients. The magnitude of the corrected anion gap correlated linearly with the anion gap (r2 = .66, corrected anion gap = 0.5 anion gap - 6.3, n = 30, p less than .01). Since the slope of the regression line for estimating corrected anion gap from anion gap was 0.5, the contribution of unmeasured anions was as important as lactate in determining the anion gap. CONCLUSIONS These data indicate that lactic acidosis does not entirely account for the metabolic acidosis during severe sepsis. Furthermore, the increased corrected anion gap suggests the presence of an unidentified anion (or anions) that is (or are) responsible, in large part, for the development of metabolic acidosis in patients with sepsis.