BACKGROUND: Tuberculosis is a highly infectious disease that is spread from person to person by infected aerosols emitted by patients with respiratory forms of the disease. We describe a novel device that utilizes immunosensor and bio-optical technology to detect M. tuberculosis antigen (Ag85B) in cough and demonstrate its use under field conditions during a pilot study in an area of high TB incidence. METHODS: The TB Breathalyzer device (Rapid Biosensor Systems Ltd) was field tested in the outpatient clinic of Adama Hospital, Ethiopia. Adults seeking diagnosis for respiratory complaints were tested. Following nebulization with 0.9% saline patients were asked to cough into a disposable collection device where cough aerosols were deposited. Devices were then inserted into a portable instrument to assess whether antigen was present in the sample. Demographic and clinical data were recorded and all patients were subjected to chest radiogram and examination of sputum by Ziehl-Nielsen microscopy. In the absence of culture treatment decisions were based on smear microscopy, chest x-ray and clinical assessment. Breathalyzer testing was undertaken by a separate physician to triage and diagnostic assessment. RESULTS: Sixty individuals were each subjected to a breathalyzer test. The procedure was well tolerated and for each patient the testing was completed in less than 10 min. Positive breath test results were recorded for 29 (48%) patients. Of 31 patients with a diagnosis of tuberculosis 23 (74%; 95% CI 55-87) were found positive for antigen in their breath and 20 (64%; 95% CI 45-80) were smear positive for acid fast bacilli in their sputum. Six patients provided apparent false positive breathalyzer results that did not correlate with a diagnosis of tuberculosis. CONCLUSIONS: We propose that the breathalyzer device described warrants further investigation as a tool for studying exhalation of M. tuberculosis. The portability, simplicity of use and speed of the test device suggest it may also find use as a tool to aid early identification of infectious cases. We recommend studies be undertaken to determine the diagnostic sensitivity and specificity of the device when compared to microbiological and clinical indicators of tuberculosis disease.