BACKGROUND: We evaluated the effects of changes in laboratory practices on the detection rates and turnaround time in a clinical mycobacteriology laboratory. METHODS: A total of 27,339 specimens from 9,947 patients were tested during the period from September 2000 to March 2004, which could be divided into the following four periods according the culture and identification methods used: Period I, use of 2% Ogawa medium for culture and niacin for identification; period II, introduction of PCR for identification; period III, introduction of BacT/Alert system (bioMerieux, Durham, USA) for culturing sterile body fluids; period IV, use of Bact/Alert system for CSF and the second of repeated sputum specimens from the same patients. During these periods, two technicians (one for the first half and the other for the second half periods) did all mycobacterial tests except PCR. RESULTS: Mean detection rates were 8.0% by auramine stain, 7.9% by nested PCR, and 6.6% by culture. The detection rates by culture for sputum specimens varied 11.4%, 7.7%, 8.1% and 5.9% in each of the four periods; for body fluids, the detection rate of 4.3% during the period III was the highest. The proportion of stain results reported within 24 hours and the identification of culture isolates within 21 days changed from 80.9% to 72.4% and from 2.3% to 30.8%, respectively. With an introduction of PCR for identification in the period II the time required for the identification of cultures decreased dramatically from 26.5 days to 4.8 days. The TAT of a direct detection by nested PCR changed from 7.2 days to 5.1 days. CONCLUSIONS: New tests should be introduced in the clinical mycobacteriology laboratory. But the cost and workload of the tests should be taken into consideration to make the laboratory service more efficient.