Eighty-four pneumococci with various MICs of penicillin (38 with MICs of less than or equal to 0.06 micrograms/ml [susceptible], 35 with MICs of 0.12 to 1.0 micrograms/ml [relatively resistant], and 11 with MICs of greater than 1.0 micrograms/ml [resistant] ) were screened by a disk diffusion test using oxacillin and methicillin to see how well they distinguished penicillin-susceptible strains from those with decreased susceptibility to penicillin. The effects of Mueller-Hinton agar plus 5% sheep blood and Trypticase soy agar plus 5% sheep blood and two atmospheres, ambient air and a candle extinction jar (increased CO2), were compared. There were no obvious differences between the effects of the two media, but zones were generally larger in ambient air than in increased CO2. Although the oxacillin test can separate penicillin-susceptible and -resistant strains, it cannot separate penicillin-resistant from relatively penicillin-resistant strains by using the breakpoint of less than 20 mm recommended by the National Committee for Clinical Laboratory Standards. When the 20-mm breakpoint was applied to methicillin, 12% of the relatively resistant strains tested were erroneously classified as susceptible. When different breakpoints were used for methicillin, there was better separation of the two classes of penicillin-resistant isolates, but a few relatively resistant strains were still classified as susceptible. We recommend that oxacillin, not methicillin, be used as the screening agent with Mueller-Hinton sheep blood agar and ambient air incubation and that the breakpoint be less than 20 mm to indicate resistance or relative resistance.