Objective To investigate the causes of false-positive rifampicin resistant results in Xpert MTB/RIF (Xpert) test for samples with extremely low bacterial loads. Methods A total of 346 samples with extremely low bacterial loads and rifampicin-resistance results from Wuhan Pulmonary Hospital between June 2017 and March 2021 were collected. The samples were divided into probe-delayed and probe dropout groups based on amplification results. Mycobacterial culture and proportion method drug susceptibility testing were performed, and Xpert retesting was conducted for strains with discordant drug susceptibility results. Results Out of the 346 samples, 195 samples (56.36%) were positive in culture. Upon Xpert retesting, among the 64 Xpert-resistant but proportion method-sensitive strains, the proportions of samples in the delayed probe group with mutations in the probe D and probe E were 4.55%（1/22） and 13.33%（2/15）, respectively. In the probe dropout group, the proportions of samples with mutations in the probe A and probe E were 75.00% (9/12) and 80.00% (8/10), respectively. The false-positive rifampicin resistance rates in the delayed probe and probe dropout groups were 78.26% (36/46) and 3.36% (5/149), respectively. Conclusion The main reasons for false-positive rifampicin resistance results in the Xpert test for samples with extremely low bacterial loads were probe delay in the D and E probes, followed by low-level drug-resistant mutations in the A and E probes.