The lack of a standardized, statistically reliable method for in vitro determinations of the minimal lethal or bactericidal concentrations of antibiotics has complicated analyses of isolates of Staphylococcus aureus which appear to be inhibited but not killed by the usual concentrations of cell wall-active antibiotics. We describe a method which identifies some of the covariants involved in determinations of minimal lethal concentrations. Lethality was defined as a 99.9% reduction in the initial inoculum of bacteria after 24 h of incubation. We limited the sample volume to 0.01 ml to minimize the inhibitory effect of antibiotic and corresponding rejection values, which detected lethality with a high degree of sensitivity and specificity. When the number of colonies on subculture was equal to or less than the rejection value, the antibiotic was considered lethal for the test organism. Rejection values encompassed initial inocula from 10(5) to 10(7) colony-forming units per ml for single and duplicate samples and allowed for 1 or 5% variability in pipette volumes and errors in initial inoculum determinations. This method was used to determine the minimal lethal concentrations of semi-synthetic penicillins for S. aureau isolates, one of which was tolerant to the killing action of penicillin.