ObjectivesWe aimed to summarise the prevalence of atypical pathogens in patients with severe pneumonia to understand the prevalence of severe pneumonia caused by atypical pathogens, improve clinical decision-making and guide antibiotic use.DesignSystematic review and meta-analysis.Data sourcesPubMed, Embase, Web of Science and Cochrane Library were searched through November 2022.Eligibility criteriaEnglish language studies enrolled consecutive cases of patients diagnosed with severe pneumonia, with complete aetiological analysis.Data extraction and synthesisWe conducted literature retrieval on PubMed, Embase, Web of Science and The Cochrane Library to estimate the prevalence ofChlamydia,MycoplasmaandLegionellain patients with severe pneumonia. After double arcsine transformation of the data, a random-effects model was used for meta-analyses to calculate the pooled prevalence of each pathogen. Meta-regression analysis was also used to explore whether the region, different diagnostic method, study population, pneumonia categories or sample size were potential sources of heterogeneity.ResultsWe included 75 eligible studies with 18 379 cases of severe pneumonia. The overall prevalence of atypical pneumonia is 8.1% (95% CI 6.3% to 10.1%) In patients with severe pneumonia, the pooled estimated prevalence ofChlamydia,MycoplasmaandLegionellawas 1.8% (95% CI 1.0% to 2.9%), 2.8% (95% CI 1.7% to 4.3%) and 4.0% (95% CI 2.8% to 5.3%), respectively. We noted significant heterogeneity in all pooled assessments. Meta-regression showed that the pneumonia category potentially influenced the prevalence rate ofChlamydia. The mean age and the diagnostic method of pathogens were likely moderators for the prevalence ofMycoplasmaandLegionella, and contribute to the heterogeneity of their prevalence.ConclusionsIn severe pneumonia, atypical pathogens are notable causes, especiallyLegionella. The diagnostic method, regional difference, sample size and other factors contribute to the heterogeneity of prevalence. The estimated prevalence and relative heterogeneity factors can help with microbiological screening, clinical treatment and future research planning.PROSPERO registration numberCRD42022373950.