AIMS--To test 10 culture media for their ability to detect resistance and sensitivity of staphylococci to methicillin by disc diffusion. METHODS--Fifty strains of Staphylococcus aureus and 135 strains of coagulase negative staphylococci were tested using Columbia, Diagnostic Sensitivity Test, Mueller Hinton, Sensitest and Iso-sensitest agars with and without 5% added sodium chloride. Cultures were examined after 18 and 40 hours of incubation. The diameter of the zone and its characteristics were recorded and these media were assessed for their ability to produce clear, readable zones of inhibition. Changes in the variables which determined resistance were investigated. Results were analysed allowing a zone diameter reduction of 8 mm and 10 mm compared with the control in addition to the standard 6 mm. RESULTS--Columbia agar with added sodium chloride supported the growth of all strains, detected the highest number of resistant strains, and was the easiest to read. Resistance was detected after 18 hours in most resistant strains, but some required 40 hours of incubation. There was poor agreement, however, on this medium (63-94%) between disc diffusion and the reference MIC method for sensitive strains. Allowing a greater reduction in zone size resulted in more agreement with sensitive strains but with consequently lowered detection of resistant strains. The other media showed some growth failures and more zones that were difficult to read. More resistance was detected when incubation was prolonged to 40 hours but this was consistently less than on Columbia agar with added salt. CONCLUSIONS--None of the media detected all of the resistant strains. Columbia agar with added salt was the most satisfactory medium in this respect, but it misidentified up to 37% of the sensitive strains as resistant. Methicillin susceptibility testing by disc diffusion testing is unreliable.