ABSTRACT: Efavirenz-based combination antiretroviral-therapy (cART) is the recommended regimen during tuberculosis (TB) therapy. In a multi-national parallel prospective-cohort study, we investigated the impact of population and pharmacogenetic variations for efavirenz pharmacokinetics, auto-induction, and immunologic outcome during antituberculosis treatment. A total of 921 treatment-naïve HIV patients with (196 Ethiopians and 231 Tanzanians) or without TB co-infection (285 Ethiopians and 209 Tanzanians) were enrolled and treated with efavirenz-based cART. TB-HIV patients started rifampicin-based anti-TB therapy 4 weeks before cART. Efavirenz plasma concentrations were measured on the 4th and 16th weeks of cART. Genotyping for CYP2B6, CYP3A5, ABCB1, UGT2B7, and SLCO1B1 was done. CD4 cells-count was measured at baseline, 12th, 24th, and 48th weeks of cART. Among HIV-only cohort, plasma efavirenz concentration and median CD4 cell count were significantly higher in Tanzanians than Ethiopians, and both CYP2B6 genotype and population-variation were significant predictors of efavirenz plasma concentration. Within-population analyses indicated a pronounced efavirenz autoinduction in Tanzanians as reflected by a significant decrease of plasma efavirenz concentration over time (p = 0.0001), but not in Ethiopians. Among TB-HIV cohort, there were no significant between-population differences in plasma efavirenz concentrations or CD4 cell-recovery, and CYP2B6 genotype but not population-variation was a significant predictor of efavirenz plasma exposure. In Tanzanian patients, short-term anti-TB co-treatment significantly reduced the mean plasma efavirenz concentration in CYP2B6*1/*1 genotype at week-4 (p = 0.005), but not at week-16 of cART. In Ethiopian patients, anti-TB cotreatment increased the mean plasma efavirenz concentration among CYP2B6*6 carriers at week-4 (p = 0.003) and week-16 (p = 0.035) of cART. In general, long-term anti-TB co-treatment increased plasma efavirenz concentration at week 16 of cART in both Ethiopians and Tanzanians being higher in CYP2B6*6/*6 > *1/*6 > *1/*1 genotypes. In TB-HIV patients, baseline body mass index (BMI), viral load, and WHO clinical-stage but not genotype, population-variation, or efavirenz concentration were significant predictors of immunologic outcome at week-48. In summary efavirenz auto-induction, pharmacokinetics, and the immunologic outcome are influenced by population-variation, anti-TB co-medication, and CYP2B6 genotype. CYP2B6 genotype is a significant predictor of efavirenz plasma exposure regardless of population-variation or antituberculosis co-treatment, but population-variation is insignificant during antituberculosis treatment. CYP2B6 genotype, population, and geographic differences need to be considered for efavirenz dosage-optimization.