Since standard susceptibility tests reflect the physiological rather than the pathological conditions prevailing within an infected abdomen, as recently documented, the effect of reduced pH and pO2 and increased pCO2 on the activity of antibiotics in vitro was studied. MICs were determined in vitro under standard culture conditions (MICstandard) and modified conditions (MICmodification) simulating the previously determined pathological values. Various classes of antibiotics were affected differently by the modified conditions. However, within an antibiotic class similar results were obtained for gram-negative and gram-positive pathogens. Median MICmodification/MICstandard ratios were 4 for aminoglycosides, 2 for quinolones and clindamycin, 1 for cephalosporins, and 0.5 for penicillins and vancomycin. Anaerobic conditions and a pH of 6.4 further increased the ratio of aminoglycosides to 8. Ratios were similar within an antibiotic class at inocula of 10(5) or 10(7) cfu/ml. All MICs determined in tests with imipenem against gram-negative and gram-positive bacteria and with vancomycin against gram-positive organisms were below the susceptibility breakpoint, whatever conditions and inocula were employed. In contrast, the percentage of MICs in susceptibility range using high inocula and modified conditions decreased to 78% for penicillins, 73% for cephalosporins, 22% for aminoglycosides, 11% for quinolones and 0% for clindamycin. In conclusion, routine susceptibility testing may overestimate the activity of aminoglycosides and underestimate the activity of beta-lactams under the conditions prevailing during abdominal infection.