Usefulness of two enzyme-linked immunosorbent assays (ELISA) for screening of dairy herds for antibodies to lipopolysaccharide (LPS) of Salmonella dublin (O:1,9,12) was investigated. Sera (3097) were collected from 40 dairy herds located in three areas of Denmark with different prevalence of salmonellosis: ten salmonellosis-free herds from the island of Samsø where there is no history of salmonellosis, ten salmonellosis-free herds from the island of Sealand where outbreaks are infrequent, and 20 salmonella infected herds from Jutland where salmonellosis is enzootic. The samples were analyzed for antibodies to S. dublin LPS using an indirect (O:9,12) and a blocking (O:9) ELISA. Using herd history of salmonellosis, herd location and clinical state of the herds as reference, the herd sensitivity and herd specificity of the tests were 100% and 100% in the indirect ELISA and 95% and 100% in the blocking ELISA, respectively. A significant correlation was found between the two tests (rs = 0.46, p < 0.001). However, the indirect ELISA detected more seropositive animals than the blocking ELISA (17% vs. 7%, respectively). In calves from Sealand, level of background reaction was significantly lower (p < 0.001) compared to the heifers and the cows. The percentages of seropositive calves in both tests were higher (p < 0.01) in comparison to cows (19 vs. 8 in indirect ELISA, and 14 vs. 6 in blocking ELISA, respectively). Results of the study indicated that it is possible to apply LPS ELISA in serological screening for salmonellosis.