Our cross-sectional study investigated the association of sub-clinical Mycobacterium avium subsp. paratuberculosis (MAP) infection with failing to produce a live offspring the season of lambing/kidding (November 2001 to January 2002) before testing (in April-May 2002), in four dairy-sheep and/or goat flocks in Greece (369 animals >or=1.5-year-old). From each selected animal 10 ml of blood and 10 g of feces from the rectum were obtained. The harvested sera were tested for antibodies to MAP with a commercial ELISA test kit; the feces were cultured on Herrold's egg-yolk medium supplemented with mycobactin J and antibiotics. An animal was considered sub-clinically infected when found either seropositive or culture positive. The true prevalence of sub-clinically infected animals, adjusted for the sensitivity and specificity of the parallel testing, was 14% (0.1-28%) and 35.9% (9.2-62.7%) in sheep and goats, respectively. The association of fertility of sheep and goats with sub-clinical paratuberculosis was investigated in random-effects logistic models. Sub-clinically infected animals (compared to uninfected) had OR for live offspring the previous year of 5.4 for parity <4, OR=0.05 for parity >6, and a non-significant OR for the middle parity category.