To assess the impact of diabetes mellitus on plasma concentration and bioavailability of rifampicin. Web of Science, Cochrane Library, PubMed and Scopus databases were screened until September 2020 on studies reported rifampicin's plasma concentration, bioavailability among diabetic tuberculosis patients and non-diabetic tuberculosis patients. According to the presence or absence of heterogeneity, the pooled estimate was operated by a random or fixed effect model. Sensitivity analysis or subgroup analysis were conducted. Attributed risk fraction of diabetes mellitus in the incidence of low rifampicin's plasma concentration 2-h after administration was calculated. Seventeen studies including 3478 tuberculosis patients were included in this study. Diabetic tuberculosis patients had 1.59 folds incidence of low rifampicin's plasma concentration 2-h after administrations (risk ratio 1.59, 95% confidence interval (1.16, 2.19), p = 0.004) and lower rifampicin's plasma concentration 2-h after administrations (mean difference -1.4, 95% confidence interval (-2.65,-0.15), p = 0.03) compared with non-diabetic tuberculosis patients. The attributed risk fraction of diabetes mellitus in the incidence of low rifampicin's plasma concentration 2-h after administration was 37%. There were no significant difference in rifampicin's maximum plasma concentration, event of low maximum plasma concentration, bioavailability and half-life between both groups. Diabetes mellitus attributed with 37% of low rifampicin's plasma concentration 2-h after administration but not influenced rifampicin's maximum plasma concentration, bioavailability and half-life. The negative effect of diabetes on plasma concentration 2hours after administration was influenced by diabetes management, income level, type of tuberculosis and its recurrence.