BACKGROUND Difficulty to obtain sputum in children complicates diagnosis of intrathoracic tuberculosis (TB). The intragastric string test (ST) used for retrieval of enteric pathogens might be an alternative specimen collection method but requires further evaluation of its utility in TB diagnosis. We conducted a cross-sectional study comparing the TB detection yield and the tolerability of ST and sputum induction (SI) in children. METHODS Two ST and SI procedures were performed in children (3-14 years of age) who were clinically suspected of having TB. The string was removed after a 2-hour gastric downtime, and SI was done after a maximum of 20 minutes nebulization with 5% saline solution. LED-fluorescence microscopy and mycobacterial cultures were performed on all specimens, and XpertMTB/RIF assay was performed on stored specimen sediments. Tolerability questionnaires were administered to parents of children. RESULTS Of 137 included children (median age: 8.1 years; 33.3% with HIV infection), 14 (10.2%) were diagnosed with TB, 10 (71.4%) by ST and 12 (85.7%) by SI. Among 105 children with both ST and SI performed, 5 (4.8%) versus 4 (3.8%) were smear positive using ST and SI, respectively (McNemar P = 1.00). Nine (8.6%) in each group had positive cultures (P = 1.00). Of 64 children tested with XpertMTB/RIF, 3 (4.7%) of the ST group versus 4 (6.3%) of the SI group were TB positive (P = 1.00). No adverse serious events were reported. ST could not be performed in 22 of 137 (16.1%) children because they were unable to swallow the capsule. CONCLUSIONS TB detection yield was comparable between ST and SI. The tolerability of ST in young children might be improved by the reduction of the size of the capsule.