Diagnosis of pulmonary tuberculosis is usually achieved by testing sputum for presence of Mycobacterium tuberculosis by microscopy, culture or nucleic acid amplification tests. However, many individuals are unable to produce sputum, particularly early in the course of illness. Studies have reported that oral swabs, assayed by nucleic acid amplification tests, may be a suitable substitute or complement to sputum testing. To determine whether this method could be useful of case finding, in which bacillary load is often lower, we evaluated it in the setting of a mass tuberculosis screening study in prison inmates in Brazil. For this sub-study, we enrolled 128 individuals with pulmonary tuberculosis confirmed by sputum Xpert testing, and 128 controls who tested negative by sputum culture and Xpert. We collected two oral swabs by participant, prior to starting treatment. Swabs were collected from the tongue by brushing along the surface for 10 times. The sensitivity of a single oral swab was 43% (N = 55/128; 95% CI: 34-52%). Using two consecutive oral swabs the sensitivity increased to 51% (N = 66/128; 95% CI: 43-60%). The specificity was 100% (128/128). In participants with high mycobaterial load in the sputum, the combined sensitivity was 90% (N = 9/10). In the participants with medium mycobaterial load in the sputum, the combined sensitivity was 79% (N = 23/29). In the participants with low or very low mycobaterial load in the sputum, the combined sensitivity was 38% (N = 34/89). Our data suggest that oral swab sampling, assayed by Xpert, has comparable sensitivity to sputum in participants with high and medium mycobacterial load in the sputum. However, 70% (89/128) of individuals identified through our mass screening study (Carbone et al.) had detection number low or very low in their sputum. In this population, oral swab testing may not have sufficient sensitivity as currently performed. Further studies are needed to identify alternative non-sputum sampling strategies in this population. © 2020 The Authors.