We investigated the potential of two different electronic noses (EN; code named "Rob" and "Walter") to differentiate between sputum headspace samples from tuberculosis (TB) patients and non-TB patients. Only samples from Ziehl-Neelsen stain (ZN)-and Mycobacterium tuberculosis culture-positive (TBPOS) sputum samples and ZN-and culture-negative (TBNEG) samples were used for headspace analysis; with EN Rob, we used 284 samples from TB suspects (56 TBPOS and 228 TBNEG samples), and with EN Walter, we used 323 samples from TB suspects (80 TBPOS and 243 TBNEG samples). The best results were obtained using advanced data extraction and linear discriminant function analysis, resulting in a sensitivity of 68%, a specificity of 69%, and an accuracy of 69% for EN Rob; for EN Walter, the results were 75%, 67%, and 69%, respectively. Further research is still required to improve the sensitivity and specificity by choosing more selective sensors and type of sampling technique.In developing countries, early diagnosis of tuberculosis (TB) followed by appropriate treatment will reduce the burden of TB. The current methods for diagnosing TB in developing countries are not satisfactory. The challenge is to develop a simple test with sensitivity at least as good as that of microscopy that can reduce the workload of the laboratory personnel, with results obtained within one working day and that can be applied when patients are in the field. We anticipate the application of new, intelligent diagnostic devices that are accessible by patients or home based for the control of disease.We have investigated the potential of two off-the-shelf electronic-nose (EN) devices (Bloodhound; Scensive Technologies Ltd., United Kingdom), named EN "Rob" and EN "Walter," in an attempt to differentiate between sputum samples from TB patients and non-TB patients in a real-life setting in Tanzania (4).An electronic nose is the colloquial name for an instrument made of an array of chemical sensors combined with some sort of pattern recognition system (1). The function of an EN is to mimic the mammalian olfactory system and produce a unique classification based on the volatile organic compounds (VOCs) present in the headspace gas being analyzed (10). In this paper, the "standard data extraction" information can be described as the maximum absorption and desorption rates during adsorption and desorption, the maximum response (or divergence), and the integrated area under the response curve; the "advanced data extraction" information is the maximum response between time points in this paper. We used principal component analysis (PCA), linear discriminant function analysis (LDA), and partial least square discriminant analysis (PLSDA) to analyze the data (9). In this paper, the odors from the headspace above sputum samples from TB patients and non-TB patients have been used. To validate the statistical software model built by the training set, we used the cross-validation, or leave-one-out, method. Sensitivity, specificity, and accuracy can then be calculated in...