A national antimicrobial resistance surveillance study was conducted from December 1997 to May 1998 to determine the prevalence of antimicrobial resistance in 6620 clinical isolates of Streptococcus pneumoniae, Haemophilus influenzae and Moraxella catarrhalis. In this centralized study, which involved 163 institutions located in 43 states, we determined MICs for representatives of five antimicrobial classes: beta-lactams (penicillin, co-amoxiclav, cefuroxime, ceftriaxone), macrolides (azithromycin, clarithromycin), co-trimoxazole, glycopeptides (vancomycin) and fluoroquinolones (levofloxacin). In most S. pneumoniae isolates, all antimicrobials were to be found active, but amongst penicillin-resistant isolates (MICs > or = 2 mg/L), resistance to other beta-lactams, macrolides and co-trimoxazole was common. For vancomycin and levofloxacin, however, activity was not associated with penicillin resistance. The prevalence of penicillin-nonsusceptible (intermediate and resistant) pneumococci was highest in the South Atlantic (44%) and East South Central (43%) regions and lowest in the Mid-Atlantic (28%) and New England (28%) regions. Resistance to beta-lactams, macrolides and co-trimoxazole was more commonly found amongst respiratory isolates than blood isolates and in strains from patients < or = 12 years old than from older patients. beta-lactamase, which was detected in 33% of H. influenzae and 92% of M. catarrhalis strains, did not affect the activity of the beta-lactams under study other than ampicillin. Certain agents, such as vancomycin and the fluoroquinolones, remain highly active, and well-designed surveillance systems that monitor MIC distributions would be needed to detect a potential for reduced susceptibility. In addition, surveillance programmes should be designed to collect information about associated resistance as well as differences in prevalence associated with region, specimen source and patient age.