Background Mycoplasma pneumoniae strains with resistance to macrolides have been spreading worldwide. Here, we aimed to clarify which antimicrobial agent is a better treatment for patients with M. pneumoniae pneumonia in a setting with large epidemics of macrolide resistance. Methods Adult patients hospitalized with laboratory-confirmed M. pneumoniae pneumonia from 2010 to 2013 were identified from the Japanese Diagnosis Procedure Combination national database. Drug switching, length of stay (LOS), 30-day mortality, and total costs for patients who underwent macrolide, quinolone, and tetracycline therapy were compared using propensity score analyses. Results Eligible patients (N = 1650) from 602 hospitals were divided into the macrolide group (n = 508), quinolone group (n = 569), or tetracycline group (n = 573). We found that 52.8%, 21.8%, and 38.6% of patients in the macrolide, quinolone, and tetracycline groups, respectively, had to switch drugs (P &lt
.0001). There was no significant difference in the LOS and the 30-day mortality rates among these 3 groups. Cost was highest in the quinolone group (P =.0062). The propensity score-matched pairs (n = 487×2) generated from the quinolone and tetracycline groups also showed a lower proportion of patients who require switches in the quinolone group than in the tetracycline group (21.2% vs 39.6%, P &lt
.0001) but not in the LOS, mortality, and cost. Conclusions There were no significant differences in the LOS and mortality among any antimycoplasmal drugs as initial treatment for hospitalized M. pneumoniae pneumonia patients despite the lower switching rate in the quinolone group.