OBJECTIVE: To assess the number of sputum specimens necessary for a reliable diagnosis of pulmonary tuberculosis (PTB) in a pastoralist population in Ethiopia. METHOD: Using routine data from Ethiopia, where three sputum specimens are currently recommended for the diagnosis of PTB, we documented, (i) the proportion of persons with suspected, PTB who submitted a first, second and third sputum specimen for smear examination and (ii) the incremental smear-positive yield from the first, to the second and third specimens. RESULTS: Of 505 persons with suspected PTB, 107 (22%) failed to submit three samples. Of 60 patients who submitted three sputum samples with at least one smear-positive sample, the first sputum sample was smear positive in 56 (93%) cases; the second sputum sample was the first to be positive in 3 (5%) cases and in only one case was the third sample the first to be smear positive (additional yield 2%). CONCLUSION: In a pastoralist setting, a reliable diagnosis of PTB can be achieved with two sputum specimens and PTB diagnosis may be adequate with just one sputum specimen. However, if this more radical approach was adopted, ways of increasing diagnostic sensitivity should be explored.