We retrospectively audited the performance of the commercial kit in use in our laboratory for the detection of Mycobacterium tuberculosis complex (MTBC) and found the sensitivity to be unacceptably low at 69% (52/75). We developed an in-house end-point polymerase chain reaction (PCR) detecting IS6110, an IS-like element of MTBC, and achieved a sensitivity of 90% (66/73) with the same DNA samples, re-emphasising the poor performance of the commercial kit. In order to avoid specificity issues surrounding gel-based PCR, we developed a probe-based real-time PCR assay with an internal control and achieved a sensitivity of 84%, specificity of 97% and diagnostic odds ratio (DOR) of 207. The evaluation was performed on clinically requested samples, so we expect the performance of the assay in real life to match the data from this evaluation. Centers for Disease Control and Prevention (CDC) guidelines recommending nucleic acid tests for the investigation of possible cases of tuberculosis are expected to promote the use of molecular assays. It is important that clinical laboratories do not assume that assays, in-house or commercial, will perform well or that they will continue to perform well. Audit at regular intervals is necessary to maintain confidence and to demonstrate that the assay works to specification in the real test population.