It has been suggested that the hydrogen ion and lactate concentrations may be superior to the polymorphonuclear cell count (PMN) in ascitic fluid, in the diagnosis of bacterial peritonitis (BP). In order to compare the diagnostic accuracy of ascitic fluid measurements of pH, lactate, glucose and the PMN in BP, we analyzed the ascitic fluids of 70 consecutive patients in whom pH, lactate, glucose and the PMN count were measured in ascitic fluid and arterial blood. Fifty-one were cirrhotic patients with uninfected ascites, 14 had BP, one tuberculous peritonitis, two ascites secondary to peritoneal metastases and two with neoplastic liver involvement but without peritoneal metastases. Statistically, highly significant differences between patients with uninfected ascitic fluid and BP were observed for ascitic fluid PMN (122 vs. 2,686 per cu mm), ascitic fluid pH (7.45 vs. 7.24), arterial-ascitic fluid pH gradient (0.02 vs. 0.22), arterial lactate (12 vs. 25 mg per dl), ascitic fluid lactate (15 vs. 45 mg per dl) and arterial-ascitic fluid lactate gradient (−3 vs. −20 mg per dl). The most reliable diagnostic cutoff levels were determined for each of the parameters: PMN > 500 per cu mm; ascitic fluid pH < 7.35; arterial-ascitic fluid pH gradient > 0.10; ascitic fluid lactate > 25 mg per dl; arterial-ascitic fluid lactate gradient < −20 mg per dl; ascitic fluid glucose < 60 mg per dl; arterial-ascitic fluid glucose gradient > 60 mg per dl. The highest sensitivity was demonstrated by the PMN count (86%), which also had the highest negative predictive value (96%). Very high specificity but lower sensitivity was demonstrated for ascitic fluid pH, arterial-ascitic fluid pH gradient, ascitic fluid lactate and the arterial-ascitic fluid lactate gradient. Diagnostic sensitivity increased when either of two parameters were required to establish the diagnosis. We believe that the most reliable combination for clinical purposes is an ascitic fluid PMN count > 500 per cu mm and/or an arterial-ascitic fluid pH gradient > 0.10. This investigation shows that an elevated PMN count in ascitic fluid is a reliable index of BP. A decreased ascitic fluid pH (<7.35) or an increased arterial-ascitic fluid pH gradient (>0.10) are about equally reliable. These tests complement each other's diagnostic ability. An ascitic fluid PMN count > 500 per cu mm and/or an arterial-ascitic fluid pH gradient greater than 0.10 are almost always diagnostic of BP, and the normality of both tests virtually excludes that diagnosis.