Some recently marketed macrolide antimicrobial agents possess physiochemical, antimicrobial, and pharmacokinetic advantages that enable their wider clinical use against Haemophilus influenzae infections. A five-laboratory study assessed the validity of existing or proposed azithromycin, clarithromycin, and erythromycin interpretive criteria for tests with H. influenzae isolates. National Committee for Clinical Laboratory Standards (NCCLS) methods, criteria, and quality-control guidelines were used. A total of 350 H. influenzae strains were processed, including fresh clinical isolates (250 strains) and replicate tests of 100 stock cultures sampling strains isolated from 1984 to 91. Azithromycin interpretive criteria (susceptible at < or = 4 micrograms/ml, > or = 12 mm) produced a 99.8% absolute agreement between the minimum inhibitory concentrations and disk diffusion results (0.2% false-susceptible error). Clarithromycin breakpoint criteria (susceptible at < or = 8 micrograms/ml, > or = 13 mm; and resistant at > or = 32 micrograms/ml, < or = 10 mm) produced high minor interpretive error, but < or = 1% combined false-susceptible and false-resistant discrepancies. Erythromycin interpretive guidelines were initially proposed for susceptible at < or = 0.5 microgram/ml, > or = 26 mm. This categorizes nearly all H. influenzae strains as resistant to this older macrolide. The NCCLS should consider the proposed erythromycin criteria for publication in appropriate tables, and a class drug should also be selected (azithromycin) that would best predict macrolide-class susceptibility for those agents indicated by the US Food and Drug Administration for H. influenzae infection chemotherapy (azithromycin and clarithromycin). No serious interpretive problems were observed with the current NCCLS criteria using Haemophilus test medium.