Mycobacterium abscessus infections tend to respond poorly to macrolide-based chemotherapy, even though the organisms appear to be susceptible to clarithromycin. Circumstantial evidence suggested that at least some M. abscessus isolates might be inducibly resistant to macrolides. Thus, the purpose of this study was to investigate the macrolide phenotype of M. abscessus clinical isolates. Inducible resistance to clarithromycin (MIC > 32 microg/ml) was found for 7 of 10 clinical isolates of M. abscessus previously considered susceptible; the remaining 3 isolates were deemed to be susceptible (MIC <or= 0.5 microg/ml). Inducible resistance was conferred by a novel erm gene, erm(41), which was present in all 10 isolates and in an isolate of Mycobacterium bolletii (M. abscessus type II). However, the erm(41) alleles were nonfunctional in the three susceptible M. abscessus isolates. No evidence of erm(41) was found in Mycobacterium chelonae, and an isolate of Mycobacterium massiliense appeared to be an erm(41) deletion mutant. Expression of erm(41) in M. abscessus conferred resistance to clarithromycin and erythromycin and the ketolide HMR3004. However, this species was found to be intrinsically resistant, independent of erm(41), to clindamycin, quinupristin (streptogramin B), and telithromycin. The ability to confer resistance to clindamycin and telithromycin, but not quinupristin, was demonstrated by expressing erm(41) in Maycobacterium smegmatis. Exposure of M. abscessus to the macrolide-lincosamide-streptogramin B-ketolide agents increased the levels of erm(41) mRNA 23- to 250-fold within 24 h. The inducible macrolide resistance phenotype of some M. abscessus isolates may explain the lack of efficacy of macrolide-based chemotherapy against this organism.