OBJECTIVE To develop a noninvasive biomarker-based detection system specific for Mycobacterium bovis for monitoring infection in wild animals. SAMPLE Serum samples from 8 experimentally infected yearling white-tailed deer (Odocoileus virginianus) and 3 age-matched control deer and from 393 Minnesota Department of Natural Resources hunter-harvested white-tailed deer in northwest Minnesota. PROCEDURES 8 yearling deer were inoculated with 2 × 10(8) CFUs of virulent M bovis strain 1315 (day 0), and sera were obtained on days 0, 19, 48, and 60; sera were obtained from 3 uninoculated control deer on those same days. Sera from these deer and 9 M bovis-positive hunter-harvested deer were tested for 3 Mycobacterium-specific biomarkers (MB1895c, MB2515c, and polyketide synthase 5) by use of an indirect ELISA. That same ELISA was used to test sera obtained from 384 exposed noninfected deer in northwest Minnesota from 2007 through 2010, concurrent with an outbreak of tuberculosis involving cattle and deer in that region. RESULTS ELISA results revealed that tuberculosis infection could be detected as early as 48 days after inoculation in experimentally infected deer. Results for 384 deer sera revealed that prevalence of tuberculosis decreased over the 4-year period. CONCLUSIONS AND CLINICAL RELEVANCE Results suggested that the prevalence of tuberculosis in Minnesota deer decreased after 2009 but tuberculosis may have persisted (as subclinical disease) at extremely low levels, as indicated by the presence of low concentrations of circulating biomarkers. Biomarker-based diagnostic tests may offer a specific approach for early identification of M bovis infection.