The expression of methicillin resistance in 10 clinical and 5 genetically constructed strains of Staphylococcus aureus has been measured in relation to temperature of incubation (37°C versus 30°C), the presence of additional sodium chloride in the medium (5.5% vs. 0.5%), and pH (7.4 vs. 5.4). Resistance was quantitatively measured by disc diffusion and agar dilution assays. In disc assays of methicillin resistance, all but one clinical isolate and both transduced strains displayed increased resistance at the lower temperature. Increased salt had little effect, or decreased the resistance of these strains. At pH 5.4 resistance decreased substantially. Methicillin resistance in three step-selected strains, by contrast, was variably affected by changes in temperature or salt concentration, though the effects were never great. The most distinctive feature of these strains was that pH change did not affect their resistance. In agar dilution assays, where surviving fractions within the population were measured as a function of antibiotic concentration, the temperature and salt effects seen in disc assays of resistance were again evident for the two clinical isolates tested, but not for the transduced strains, where added salt increased resistance at the higher concentrations of methicillin tested. For two step-selected strains, the effects of lowering temperature and increasing salt on resistance were variable. By contrast with the disc assays, these strains were less resistant at the lower pH, though not to the same relative extent as that shown by the clinical isolates and transduced strains. Resistance heterogeneity was not significantly affected by higher salt and lower incubation temperature in the two clinical strains tested, but elevated salt consistently increased heterogeneity in transduced and step-selected strains at the higher concentrations of methicillin tested. In the absence of salt, step-selected strains exhibited little heterogeneity, unlike the clinical and transduced strains. We conclude that the similarities and differences seen in clinical isolates and laboratory strains favour the idea that methicillin resistance arises clinically in S. aureus as a result of transduction (or other genetic transfer) of existing gene(s) determining resistance, rather than as a result of selection of mutants among sensitive or less resistant strains. Our findings also indicate that inclusion of salt in media containing methicillin may attenuate resistance to the antibiotic and thus decrease the sensitivity of detection of marginally resistant strains of S. aureus.