A sample survey with the objective of determining the prevalence of bovine tuberculosis by means of an intradermal tuberculin test was conducted in Madagascar and it was found that the prevalence rate varied from 0-30% by veterinary district. In order to estimate the true prevalence, the validity of the test was investigated by assessing its sensitivity and specificity in two groups of animals from two different regions, which were destined for slaughter. In the first group where the probability of non-infected animals should have been the highest, sensitivity was estimated at 0.52 (n = 21) and specificity at 0.99 (n = 79). In the second group selected on the basis of apparent ill health of the animals in a high-prevalence bovine tuberculosis area, sensitivity was estimated at 0.8 (n = 10) and specificity at 1 (n = 12). The results obtained from both groups of cattle were not combined for statistical purposes because the sensitivity of the skin test seemed to fluctuate in relation to the chronicity of the disease. These fluctuations are discussed. However, since the first group of zebu cattle was more representative of the cattle population across the country as a whole, its results were retained as operational parameters for further screening.