Macrolide-resistant Propionibacterium acnes are frequently isolated from patients with acne vulgaris, and the most resistant isolates (>90% resistance) have the 23S rRNA mutation. An increase in resistant P. acnes with this mutation is thought to be caused by the inappropriate use of antimicrobials. Therefore, we studied the mutation frequency of macrolide resistance in P. acnes in vitro. When P. acnes mutants were exposed to clarithromycin after being incubated in broth without antimicrobials, resistant mutants with the 23S rRNA mutation were not isolated. However, the mutants were obtained at the frequency of 10 after being pre-incubated with 0.03 μg/mL of antimicrobials. This is the estimated epidermal concentration of clarithromycin after p.o. administration. The resistant mutants had the 23S rRNA mutations A2058G, A2059G and C2611G. When pre-incubated with clarithromycin, C2611G mutants which showed resistance to clarithromycin were obtained 32.1% more often than pre-incubated with clindamycin (P < 0.01). By contrast, when pre-incubated with clindamycin, A2058G mutants, which show high-level resistance to both clarithromycin and clindamycin, were more frequently obtained than pre-incubated with clarithromycin (87.5%, P < 0.01). No difference in the isolation rate of A2059G mutants, which show high-level resistance to macrolides but low-level resistance to clindamycin, was found with either treatment. These results indicate the possibility that long-term use of oral macrolides for acne treatment facilitate the increase of macrolide-resistant P. acnes.