Despite the longstanding presence of grass carp Ctenopharyngodon idella in the Upper Mississippi River (UMR) watershed, information regarding their populations remains largely unknown, in part because capture is difficult. Occupancy models are a popular wildlife assessment tool to account for imperfect detections but have been slow to be adopted in fisheries. Herein, we used occupancy modelling to evaluate the influence of two environmental covariates (river discharge and water temperature) on grass carp occupancy, extinction, colonization, and detection at nine sites within south-eastern Iowa rivers from April to October 2014 and 2015. Grass carp were detected at least once at all but one site. The most parsimonious model indicated that grass carp colonization probability increased from 0.15 to 0.67 with increases in river discharge. In contrast, occupancy (0.20), extinction (0.29), and detection (0.50) probabilities were temporally constant. Models indicated that water temperatures did not influence grass carp extinction or colonization probabilities relative to river discharge. Cumulative grass carp detection probability approached 1.0, whereas conditional occupancy estimates were less than 0.1 when using five or more sampling transects. The use of a robust design occupancy model allowed us to estimate site occupancy rates of grass carp corrected for imperfect detections, while demonstrating the importance of river discharge for site colonization. These results can be used to assess the distribution of a cryptic fish while helping to guide grass carp sampling and removal efforts.