The Florida east coast terrapin (Malaclemys terrapin tequesta) is a rare and potentially endangered species that is difficult to survey because of poor detection probability and a patchy distribution. Like many rare species sampling programs, we apply a density and multistate occupancy sampling approach that considered the impacts of imperfect detection. We separated the detection process into availability of the animal within the sampling area (e.g., coming to surface) and perceptibility (actually seeing it). Our study employed a density estimation approach originally developed for birds, which combined time to detection and distance sampling within a Bayesian N-mixture model. Our study also estimated two abundance occupancy states (few and many terrapins). We were able to estimate large differences in terrapin densities between sites through functions of site-specific environmental covariates (water depth, distance to mangrove, and distance to land). The detection probability was poor (0.28) for the few terrapin occupancy state, but was much greater for many (0.75). The time to detection and distance-sampling approach for this aquatic animal should be useful for other aquatic organisms that regularly surface. Terrapins were generally available to be sighted within four minutes but detection declined rapidly to a low probability of terrapin detection >30 m. The approach of estimating density as a function of habitat covariates to identify habitat associations can provide an effective method that combined with adaptive sampling should be useful to investigate the distribution of terrapins in open water.