Testing for Mycobacterium avium subsp. paratuberculosis (MAP) is typically done on a group of cows (e.g. in control programs), which changes the interpretation of the herd test results. In New York State, a kinetics ELISA (KELA) with multiple thresholds is used and the results of individual cows are reported to the farmers. The implication of the KELA results for the whole herd are not reported yet. In a previous study, the predictive value of the KELA for moderate or heavy shedding was obtained at cow level. The objective of the current study was to use probability theory to interpret KELA results at the herd-level. The likelihood of being a non-infected or an infected herd was determined based on the proportion of cows in the four KELA categories. Moreover, for infected herds the expected true prevalence was estimated. The herd-level interpretation of the test results was validated on real herd data. The probability that all cows test negative (KELA<65) decreases rapidly from 95% when a single cow is tested to 5% when 60 cows were sampled. When in a sample of 10 cows a single cow tests positive (65<KELA90%) to contain at least one truly infected cow. Herd-profiles for the probability of infection were developed based on the distribution of cows in the KELA categories. The probability for a herd to be infected is mostly dependent on the proportion of cows in the highest KELA category.