The aim of this work was to investigate whether an enzyme-linked immunosorbent assay (ELISA) was useful for early detection of maedi-visna virus (MVV) infection in sheep under field conditions. An ELISA based on p25 recombinant protein and a gp46 synthetic peptide was used. Sequentially obtained serum samples (n = 1,941) were studied for 4 years. ELISA results were compared with those of the agar gel immunodiffusion (AGID) test, and results of both tests were compared with a reference result established using consensus scores for at least 2 of 3 serologic techniques (AGID, ELISA, and western blotting, which was used to resolve result discrepancies between the other 2 techniques). A total of 247 discrepancies were observed between ELISA and AGID. Of these, 131 were due to an earlier detection of 120 sera by the ELISA and 11 sera by AGID. The remaining discrepancies (116) were due to the presence of false reactions in both tests. Fewer false-negative results were found by ELISA than with AGID (6 vs. 69 sera, respectively), whereas the number of false-positive results was virtually the same for ELISA and AGID (21 vs. 20, respectively). In relation to the reference result, ELISA sensitivity and specificity were 97.8% and 98.2%, respectively, whereas values for AGID were 76.3% and 98.3%, respectively. The agreement between ELISA and the reference result was higher than that between AGID and the reference result (K value: 0.96 and 0.77, respectively). A variation in the ELISA signal (based on optical density) was observed during the study period, suggesting different antibody levels throughout the animal's life. The ELISA was useful for detecting MVV-infected sheep in field conditions and has potential for use in control and eradication programs.