Small ruminant lentivirus (SRLV) control programs are mainly based on diagnostic tests performed on blood samples collected from sheep and goats. Since blood sampling is costly and stressful for the animals, we evaluated whether milk could be used as an inexpensive and easily collectable matrix for SRLV detection. We therefore compared SRLV detection via two commercial enzyme-linked immunosorbent assays (ELISAs) and quantitative polymerase chain reaction (qPCR) in blood and corresponding milk samples from 321 goats originating from eight different SRLV-infected farms in Flanders (Belgium). The IDscreen ELISA had a better relative sensitivity (97% vs 93%) and specificity (100% and 97%) than the Elitest ELISA for SRLV-specific antibody detection in milk compared to serum. The higher sensitivity correlates with a 10-fold higher analytical sensitivity of the IDscreen test. In contrast to the overall good ELISA results, qPCR on milk cell pellets lacked sensitivity (81%) and specificity (88%), compared to molecular detection in blood leucocyte pellets. Our results show that serology is more suitable than qPCR for SRLV diagnosis, and that milk may represent an interesting matrix for a preliminary evaluation of a herd's infection status. Serum remains however the sample of choice for control programs where it is important to identify positive animals with the highest sensitivity.