Considering the poor facilities available for microbiological diagnosis in some countries where Brucella abortus is a frequent cause of bovine abortion, a study was conducted to determine if isolation of B. abortus from an aborted bovine fetus could be predicted from a detailed histological study of the formalized lung. Thirty-nine samples of B. abortus positive and 20 negative fetal samples were examined for the presence of 14 different pulmonary lesions. Differences in the frequency of observed lesions between the positive and negative groups, were determined by odds ratios and chi square statistic. The confidence of the prediction was calculated by means of the logistic computer model. The frequency of eight lung lesions was found to be significantly (p less than 0.05) different between the groups; nevertheless, these lesions were not specific enough to be able to incriminate B. abortus as the cause of abortion.