The multilocus short sequence repeat (MLSSR) method is a newly described approach for genotyping Mycobacterium avium subsp. paratuberculosis (M. paratuberculosis) that has been used to differentiate strains previously indistinguishable by other methods. This study used four of the most divergent MLSSR loci (L1, L2, L8 and L9) to survey a total of 149 different strains of M. paratuberculosis from dairy cattle across the United States. Loci were sequenced multiple times to ensure consistency. Both animal (n = 138) and environmental (n = 23) sources from 29 dairy herds in 17 states were represented in this collection. The Simpson's index of diversity was 0.73, 0.80, 0.30, and 0.34 for each of the four loci, respectively. Temporal stability studies of these loci are currently underway. Overall, the 149 M. paratuberculosis strains could be differentiated into 53 different genotypes, but no geographic correlation within the subtypes was observed. Multiple genotypes were recovered from 24 of the 29 herds represented by these strains. The same genotype was recovered from two other herds, and the final three herds were represented by a single isolate. Of the 13 herds from which both animal and environmental isolates were obtained, 6 had at least one animal and environmental isolate with the same genotype. To determine if multiple strains could be recovered from the same animal, two dairy cows naturally infected with M. paratuberculosis and housed in Biosafety Level III containment at NVSL were used. Six isolates apiece were recovered from these animals over the course of ten months. For both animals, five different genotypes of M. paratuberculosis were recovered. These results taken together suggest that multiple strains of M. paratuberculosis observed within a herd may be the result of individual animals being infected with more than one strain of this pathogen.