Background: the national and international guidelines of the WHO and the IUALTD recommend priority sputum smears for diagnosis of pulmonary tuberculosis. We studied the use and diagnostic yield of sputum, broncho-alveolar and gastric lavage (BAL, GL) in the reference national mycobacteria Laboratory (NRL) (Burkina Faso) to see if clinicians follow the recommendations in the prescription of specimens. Methods: We analyzed by microscopy all samples sent to the RNL from 8 public and 6 private health care facilities in 2010. Results: 2638 patients were included. We have reported on 2414 early morning sputum samples, BAL from 191 patients and gastric lavage from 33. There were significant differences in the type of sample submitted per institution. Sputum formed 95% of the 862 samples from one hospital, whereas 78% of the 223 samples submitted by one private clinic were BAL samples. No sputum samples were submitted from these patients and in this clinic the diagnostic yield of BAL was just 4%. The overall diagnostic yield for early morning sputum specimens was 12% (290/2414), for BAL 3.7% (7/191) and for GL 6% (2/33). Conclusion: there was considerable variation in the diagnostic procedure used. Invasive methods are used at some sites in preference to safer and simpler means of obtaining material for diagnosis. Our review shows the need for constant vigilance and close adherence to internationally accepted guidelines for the diagnosis of pulmonary TB.