Background Tuberculosis (TB) is a major public health problem in the Horn of Africa with Ethiopia being the most affected where TB cases increase at the rate of 2.6% each year. One of the main contributing factors for this rise is increasing transmission due to large number of untreated patients, serving as reservoirs of the infection within the communities. Reduction of the time between onset of TB symptoms to diagnosis is therefore a prerequisite to bring the TB epidemic under control. The aim of this study was to measure duration of delay among pastoralist TB patients at TB management units in Somali Regional State (SRS) of Ethiopia. Methods A cross sectional study of 226 TB patients with pastoralist identity was conducted in SRS of Ethiopia from June to September 2007. Patients were interviewed using questionnaire based interview. Time between onset of TB symptoms and first visit to a professional health care provider (patient delay), and the time between first visits to the professional health care provider to the date of diagnosis (medical provider's delay) were analyzed. Both pulmonary and extrapulmonary TB patients were included in the study. Result A total of 226 pastoralist TB patients were included in this study; 93 (41.2%) were nomadic pastoralists and 133 (58.8%) were agro-pastoralists. Median patient delay was found to be 60 days with range of 10–1800 days (83 days for nomadic pastoralists and 57 days for agro-pastoralists). Median health care provider's delay was 6 days and median total delay was 70 days in this study. Patient delay constituted 86% of the total delay. In multivariate logistic regression analysis, nomadic pastoralism (aOR. 2.69, CI 1.47–4.91) and having low biomedical knowledge on TB (aOR. 2.02, CI 1.02–3.98) were significantly associated with prolonged patient delay. However, the only observed risk factor for very long patient delay >120 days was distance to health facility (aOR.4.23, CI 1.32–13.54). Extra-pulmonary TB was the only observed predictor for health care providers' delay (aOR. 3.39, CI 1.68–6.83). Conclusion Patient delay observed among pastoralist TB patients in SRS is one of the highest reported so far from developing countries, exceeding two years in some patients. This long patient delay appears to be associated with patient's inadequate knowledge of the disease and distance to health care facility with nomadic pastoralists being the most affected. Regional TB control programmes need to consider the exceptional circumstances of pastoralists, to maximise their access to TB services.