<sec> <title>BACKGROUND:</title> The use of molecular amplification as-says for TB diagnosis is limited by their costs and cartridge stocks. Pooling multiple samples to test them together is reported to have similar accuracy to individual testing and to save costs. </sec> <sec> <title>METHODS:</title> Two surveys of individuals with presumptive TB were conducted to assess the performance of pooled testing using Xpert® MTB/RIF (MTB/RIF) and Xpert® Ultra (Ultra). </sec> <sec> <title>RESULTS:</title> A total of 500 individuals were tested using MTB/RIF, with 72 (14.4%) being MTB-positive. The samples were tested in 125 pools, with 50 pools having 1 MTB-positive and 75 only MTB-negative samples: 46/50 (92%, 95% CI 80.8–97.8) MTB-positive pools tested MTB-positive and 71/75 (94.7%, 95% CI 86.9–98.5) MTB-negative pools tested MTB-negative in the pooled test (agreement: 93.6%, κ = 0.867). Five hundred additional samples were tested using Ultra, with 60 (12%) being MTB-positive. Samples were tested in 125 pools, with 42 having 1 MTB-positive and 83 only MTB-negative samples: 35/42 (83.6%, 95% CI 68.6–93.0) MTB-positive pools tested MTB-positive and 82/83 (98.8%, 95% CI 93.5–100.0) MTB-negative pools tested MTB-negative in the pooled test (agreement: 93.6%, κ = 0.851; P > 0.1 between individual and pooled testing). Pooled testing saved 35% (MTB/RIF) and 46% (Ultra) of cartridges. </sec> <sec> <title>CONCLUSIONS:</title> Pooled and individual testing has a high level of agreement and improves testing efficiency. </sec>