Serologic and fecal culture assays for paratuberculosis are being used on an increasing basis to identify infected cattle herds and control Johne's disease. Despite this, the utility of these assays in cattle herds for these purposes is not well understood. The objective of this study was to compare results from a serum ELISA and fecal culture in infected and uninfected dairy cattle herds, in order to define optimal use of these tests in herd control programs. Fecal and serum samples from all adult cows in 49 infected herds and 7 uninfected herds were tested using fecal culture and a commercially available ELISA in Minnesota and Pennsylvania. Results from this study indicated that the specificity of the ELISA in uninfected dairy herds varied by herds selected, 97% in one group of 4 herds and 72% in another 3 herds. Within infected dairy cattle herds, 40% of culture-positive cows were ELISA-positive, an indication of the relative sensitivity of ELISA compared to detectable fecal shedding, and similar to the ELISA sensitivity estimates reported in the scientific literature. ELISA relative sensitivity, however, varied by fecal shedding prevalence from 57% in herds with <5% shedding prevalence to 33% in herds with greater than 15% shedding prevalence. This indicates different interpretation should be applied to test results from dairy herds with low compared to high fecal shedding prevalence. In infected dairy herds with <5% shedding prevalence, fecal shedding was detected in 33% of cows with ELISA S/P above 1.0, compared to 88% of cows with ELISA S/P above 1.0 in herds with greater than 15% shedding prevalence.