OBJECTIVE To compare the prognostic value of blood lactate concentrations, gastric intramucosal pH, and their combination in patients with severe sepsis. METHODS Prospective, noninterventional study. METHODS Medical/surgical intensive care unit of a university hospital. METHODS The study included 35 consecutive patients (44 to 82 yrs) with severe sepsis as defined by fever or hypothermia (rectal temperature > 38.3 degrees or < 35.5 degrees C), tachycardia (heart rate > 100 beats/min), tachypnea (respiratory rate > 20 breaths/min) or mechanical ventilation, abnormal white blood cell count (> 10 or < 6 x 10(3) cells/mm3), hypotension (systolic arterial pressure < 90 mm Hg), and evidence of organ dysfunction (oliguria or deterioration of mental status). METHODS Arterial lactate concentration and intramucosal pH were measured at the time of study entry, and at 4 and 24 hrs later. Hemodynamic data and oxygen-derived variables were determined at the time of study entry and 24 hrs later. Arterial blood and balloon saline gases were also determined to obtain the pH gap (arterial pH-intramucosal pH) and the PCO2 gap (intramural PCO2-PaCO2). RESULTS Of the 35 patients, 19 survived the intensive care unit stay. At the time of study admission, 23 (66%) patients had an increased lactate concentration (> 2 mEq/L) and 26 (74%) had a low intramucosal pH (< 7.32). Initially, there were no significant differences in blood lactate concentrations between nonsurvivors and survivors (3.2 +/- 1.5 vs. 2.8 +/- 2.3 mEq/L). Lactate concentrations remained high in nonsurvivors and progressively decreased in survivors (4 hrs: 3.3 +/- 1.1 mEq/L in nonsurvivors vs. 2.2 +/- 0.9 mEq/L in survivors [p < .01]; 24 hrs: 3.5 +/- 2.0 mEq/L in nonsurvivors vs. 1.9 +/- 1.1 mEq/L in survivors [p < .05]). Intramucosal pH was lower in the nonsurvivors than in the survivors initially (7.19 +/- 0.15 in nonsurvivors vs. 7.30 +/- 0.14 in survivors [p < .05]), at 4 hrs (7.18 +/- 0.17 in nonsurvivors vs. 7.29 +/- 0.13 in survivors [p = .06]), and at 24 hrs (7.19 +/- 0.31 in nonsurvivors vs. 7.30 +/- 0.17 in survivors [p < .05]). Of the 23 patients with initially high lactate concentrations, 12 (60%) of the 20 patients with low intramucosal pH died, as compared with one (33%) of the three patients with normal intramucosal pH (p = .052). Of the 14 patients with persistently high lactate concentrations at 24 hrs, all nine (100%) patients with low intramucosal pH, but only two (40%) of five patients with normal intramucosal pH died (p < .001). No significant relationship was found between lactate or intramucosal pH and oxygen-derived variables. Intramucosal pH correlated better with gastric intramural PCO2 (r2 = .58) than with arterial bicarbonate or base deficit/excess. Intramural PCO2 was a more specific predictor of mortality than intramucosal pH. When compared with patients with normal lactate concentrations, those patients with high lactate concentrations had a higher pH gap (0.22 +/- 0.22 vs. 0.07 +/- 0.13 [p < .01]) and PCO2 gap [21.0 +/- 33.9 vs. 1.8 +/- 9.8 torr [2.79 +/- 4.5 vs. 0.24 +/- 1.34 kPa]; p < .01). CONCLUSIONS Both lactate concentrations and intramucosal pH represent reliable prognostic indicators in severe sepsis, and their combination improves the prognostic assessment in these patients. Both variables are better prognostic indicators than oxygen-derived variables. Intramural PCO2 appears to be a more specific variable than intramucosal pH, which partially reflects systemic metabolic acidosis. Combined determinations of blood lactate concentrations and intramucosal pH or intramural PCO2 may help to predict outcome from severe sepsis.