ELISA tests for detection of antibody to M. paratuberculosis usually lack sensitivity and/or specificity. These deficits are attributable to anergy in recently infected cattle and cross reactivities of antibodies for antigens used in the assays. When specificity is set at about 99%, most ELISAs have a sensitivity ranging from about 40-60%. Of 473 fecal culture-positive animals, only about 40% were detected by an ELISA having a 99% specificity. Three-hundred nineteen of these animals (67%) were low-shedders (less than 10 colonies per culture); only 21% of low-shedders were seropositive. Alternatively, 88 of 94 (94%) high-shedders (greater than 100 colonies per culture) and 58% of moderate shedders (10-100 colonies) were antibody-positive. Because low-shedders account for about two-thirds of infected animals, conventional ELISAs miss many infected cattle. We concluded that if ELISA were to be an effective tool to determine herd seroprevalence, classification of animals as positive or negative based upon a single threshold in ELISA would lead to an unacceptable number of false negative results. Using 1340 samples from farms of known Johne's disease status, we established three arbitrary thresholds for ELISA: the first was 2 times the mean ELISA value for all uninfected animals (n = 867), the second was 3.5 times the mean of all negatives, and the third was equivalent to the 99% specificity level which was 10.2 times the mean ELISA for uninfected cattle. These arbitrary thresholds defined 4 categories of ELISA results. If ELISA values were in the first category, the animals were classified as being at low risk of developing Johne's disease; if in the second, third, or fourth categories, the risk was considered moderate, moderately-high, or high, respectively. A herd profile was then established based on the proportions of animals that fell into the various risk categories. For example, a herd with 95% of the animals in the low- risk and 5% in the moderate-risk categories virtually always depicted a Johne's-free herd, as determined by repeated sequential fecal test. Conversely, if 25% of the animals were in the moderately-high to high-risk categories, such a herd inevitably was infected based on fecal culture confirmation. We conclude that profiling herds via a multi-threshold, rather than a single-threshold ELISA, ia a more accurate means of determining the paratuberculosis herd infection status.