This study investigated the microsite, patch, and landscape level habitat requirements of the phlox moth (Schinia Indiana, Noctuidae). Detection-nondetection and habitat data were collected at a total of 121 host-plant patches spread across 60 km2 at the Fort McCoy military installation in Wisconsin, USA. The relationships between the probability of patch occupancy and the patch and landscape variables were modeled using Program PRESENCE. A subset of patches was resampled to estimate the probability of detecting a moth when present, given imperfect detection. The subset of models which had the best predictive abilities were compared using a suite of model diagnostics including prediction error using the training set-test set approach, Hosmer-Lemeshow goodness of fit tests, and assessment of the standardized residuals. Patch size, distance to the nearest previously occupied patch, and the average number of buds per phlox stem within a patch were the most important predictors of moth occupancy, while distance to the nearest patch, bordering road type, and the average number of blooms per stem were less important predictors. Cloud cover and precipitation influenced the probability of detecting moths during surveys. Host-plant stem spacing, canopy cover, size of the nearest patch, and the total area of patches on the surrounding landscape were not found to be important predictors of moth occupancy in this study. Microsite habitat preferences were examined using MANOVA. At the microsite level, moths were found more often stems with large numbers of buds and low canopy cover, located in areas with high stem density. Moths were not found more frequently in microsites with greater percentages of stems in bloom, or for stems with greater numbers of blooms than randomly selected stems. Management recommendations include increasing patch sizes and connectivity, and mitigating habitat loss by replacing patches with patches equal or greater in occupancy probability, rather than simply considering patch size.