The activities of ampicillin-sulbactam and amoxicillin-clavulanate were studied with 100 selected clinical Escherichia coli isolates with different beta-lactam susceptibility phenotypes by standard agar dilution and disk diffusion techniques and with a commercial microdilution system (PASCO). A fixed ratio (2:1) and a fixed concentration (clavulanate, 2 and 4 micrograms/ml; sulbactam, 8 micrograms/ml) were used in the agar dilution technique. The resistance frequencies for amoxicillin-clavulanate with different techniques were as follows: fixed ratio agar dilution, 12%; fixed concentration 4-micrograms/ml agar dilution, 17%; fixed ratio microdilution, 9%; and disk diffusion, 9%. Marked discrepancies were found when these results were compared with those obtained with ampicillin-sulbactam (26 to 52% resistance), showing that susceptibility to amoxicillin-clavulanic acid cannot be predicted by testing the isolate against ampicillin-sulbactam. Interestingly, the discrimination between susceptible and intermediate isolates was better achieved with 4 micrograms of clavulanate per ml than with the fixed ratio. In contrast, amoxicillin susceptibility was not sufficiently restored when 2 micrograms of clavulanate per ml was used, particularly in moderate (mean beta-lactamase activity, 50.8 mU/mg of protein) and high-level (215 mU/mg) TEM-1 beta-lactamase producer isolates. Four micrograms of clavulanate per milliliter could be a reasonable alternative to the 2:1 fixed ratio, because most high-level beta-lactamase-hyperproducing isolates would be categorized as nonsusceptible, and low- and moderate-level beta-lactamase-producing isolates would be categorized as nonresistant. This approach cannot be applied to sulbactam, either with the fixed 2:1 ratio or with the 8-micrograms/ml fixed concentration, because many low-level beta-lactamase-producing isolates would be classified in the resistant category. These findings call for a review of breakpoints for beta-lactam-beta-lactamase inhibitors combinations.