Three i-ELISAs using LPS, the immunodominant component of Brucella abortus, were developed with three different conjugates: monoclonal antibodies 1C8 (anti-bovine IgG(1)) and 3H3 (mainly specific for bovine IgG(2) but also reacting with IgG(1)) and protein G (reacts with both bovine IgG subclasses). Using a cut-off value of 2.5U/ml, the i-ELISA with 3H3 as conjugate had a specificity (95% CI: 98.32-99.63%) that was significantly higher than the same assay with 1C8 (95% CI: 96.08-98.26%) or PG (95% CI: 95.83-98.09%). In areas where false positive serological reactions (FPSR) were common, the specificity of the i-ELISAs decreased significantly. The specificity of the i-ELISAs increased with the age of the animals tested, irrespective of the conjugate. The specificity of the i-ELISAs and traditional tests was also examined using sera from animals infected per os with bacteria bearing LPS similar to the Brucella LPS. It appeared that Yersinia enterocolitica O:9, Xanthomonas maltophilia and Salmonella urbana infections induced FPSR both in the i-ELISAs and in the traditional tests, but the 3H3 assay was significantly less prone to produce false positive reactions than the 1C8 and PG assays. The i-ELISAs were more sensitive, allowed earlier detection, and were more persistent than the traditional serological tests both in experimentally and naturally Brucella-infected animals. Weekly i-ELISA monitoring of experimentally infected pregnant heifers (previously vaccinated or not) allowed a prediction of abortion. Furthermore, the 1C8 assay showed significantly higher titres irrespective of day post-infection and vaccination status. The accuracy of the assay could be improved by making use of additional information (e.g. animal age or conjugate) and by selecting appropriate cut-off points on the basis of the prevailing epidemiological situation. The i-ELISAs appear an appropriate choice in order to maintain an official brucellosis-free status because of their sensitivity, early detection and long persistence and, for the same reasons, seem especially valuable for the detection of latent carriers (i.e. animals classified negative by classical serological tests) among imported animals.