A routine disc diffusion method for detecting antibiotic-resistant Streptococcus pneumoniae was evaluated in 20 clinical microbiology laboratories. Fifty consecutive clinical isolates were tested by disc diffusion in each laboratory, and collected for MIC determinations. Participating laboratories used three disc types and several media; altogether 17 disc-medium-breakpoint reference combinations were used. Of the 781 isolates, 1.2% were resistant to penicillin and 4.2% were intermediately resistant. Eight of the nine resistant isolates had been classified as resistant to penicillin using a 1 microgram oxacillin disc screening test in the participating laboratories. If the MIC was taken as 0.125 mg/L, which is just above the intermediate breakpoint, 11 of 12 isolates were interpreted as susceptible to penicillin with the oxacillin disc test. The laboratories using Oxoid or Biodisk paper discs performed better in detecting co-trimoxazole resistance than the laboratories using Rosco tablets. Two control strains (one multiresistant and one susceptible) were each tested in the laboratories ten times. Based on these results laboratory-specific breakpoints for tetracycline were determined linearly. These adjusted breakpoints corrected three of five false-susceptible interpretations of the resistant clinical isolates and reduced the number of minor errors from 8.5% to 2.6%.