The wild white-tailed deer (Odocoileus virginianus) population in Michigan, USA, has endemic Mycobacterium bovis. We determined whether there were spatial clusters of retrospective TB cases in white-tailed deer in northeastern Michigan and identified specific factors associated with the spatial clusters. Data from hunter-harvested deer (age, gender, TB status, and geographic section) were collected by the Michigan Department of Natural Resources (MDNR) during TB surveillance from 1995 to 2002. Land cover (vegetation, land-use) and land type (soil types and drainage characteristics, landforms) described potential deer habitats. Specific locations of large-scale supplemental feeding sites were collected from the MDNR aerial surveillance program from 1997 to 2002. Analyses were conducted using principal components derived from environmental data (and other risk factors) on spatial clusters of disease (identified by the spatial scan statistic). Spatial effects were incorporated into the multivariable analyses by using a neighborhood approach. A total of 420 deer with M. bovis infection were identified from 1995 to 2002, out of 39,451 harvested deer from 3216 TRS units, and spatial clusters of cases were identified. A total of seven principal components of environmental data were generated. Clusters were associated with the presence of large expanses of deciduous forests on moraine ridges separated by low areas of forested wetlands, and the presence of many small lakes. Factors that promoted congregation of deer for extended periods of time (natural cover, access to water, and less human contact) appeared to be associated with increased odds of TB positivity. This suggests that there are specific areas where interventions can be implemented to reduce congregation of animals and disrupt the cycle of infection transmission.