The benefit of empiric coverage for community-acquired pneumonia (CAP) for atypical bacteria is controversial. This meta-analysis purpose was to compare the clinical failure rate between adults who empirically received atypical coverage versus those who did not. We searched PubMed and EMBASE for randomized controlled trials (RCTs), comparing the clinical failure rate of CAP associated with individual atypical bacteria between adults who received empiric atypical coverage versus those who did not. Risk differences (RDs) with 95% confidence intervals (CIs) were calculated using random-effects models. Eight double-blind RCTs (65 patients with Legionella spp., 176 patients with M. pneumoniae, and 78 patients with C. pneumoniae) were included in the meta-analysis. The rate of clinical failure was significantly lower with empiric atypical coverage in CAP associated with Legionella spp. (RD, −42.6%; 95% CI, −69.8% to −15.4%; p-value = 0.002; I2 = 0%) and Mycoplasma pneumoniae (RD, −9.5%; 95% CI, −18.9% to −0.1%; p-value = 0.048; I2 = 0%), but not with Chlamydia pneumoniae (RD, 7.1%; 95% CI, −9.0% to 23.1%; p-value = 0.390; I2 = 0%). This meta-analysis of RCTs found that empiric atypical coverage decreased the clinical failure rate of CAP associated with Legionella spp. and M. pneumoniae, but not with C. pneumoniae.