The complement fixation test (CFT), the enzyme labelled immunosorbent assay (ELISA) and the gel diffusion precipitin test (GD) were compared, for the diagnosis of Brucellu ovis infection in rams. The sensitivities of the tests in 109 rams which were shedding B. ovis in their semen were: CFT 96.3%; ELISA 97.2%; GD 91.7%. The specificities of the tests in 141 rams from non-infected flocks were: CFI 99.3%; ELISA 98.6%; GD 100%. Predictive values of the three tests were measured in 285 rams from infected flocks. Thirty-eight percent of these rams were shedding B. ovis in their semen. Predictive values of positive tests were: CFT 75.5%; ELISA 66.7%; GD 72.5%. Predictive values of negative tests were: CFI 97.1%; ELISA 97.6%; GD 93.8%.

The importance of correct interpretation of serological test results, common sources of error and problems associated with tests are discussed. In bovine brucellosis, a disease which is ideally suited to serological diagnosis, foetal contact with infection may cause the calf to be a serologically negative carrier. The immune tolerant animal resulting from foetal contact with virus is a major problem in the serological detection of border disease. In Johnes' disease and to a lesser extent in Brucella ovis and leptospiral infections, problems associated with sensitivity and specificity of the tests are stressed. Serovar specificity, cumbersome test procedures and negative tests in the incubation period cause difficulty in the interpretation of serological test results for leptospirosis. The importance of clinical examination, herd histories and alternative diagnostic procedures is important in all diseases. Wherever possible, flocks or herds should be maintained in specific disease-free state. Selection of stock from accredited herds or flocks is the most certain method of buying non-infected animals.