The efficiency of survey gear is an important measure that can be used to estimate the absolute scale of populations in assessment models. We develop a general hierarchical model for estimating the efficiency of a New Bedford-style sea scallop (Placopecten magellanicus) survey dredge from paired dredge and Habcam camera system tows. Habcam data for each tow consist of multiple images that give information on within-tow variability of scallop density. The model accounts for these multiple observations as well as the possibility of differences between the true densities in dredge and Habcam pairs due to the pairs not covering exactly the same ground. We fit several models with alternative assumptions to observations on Atlantic sea scallops and compare the relative performance using Akaike’s information criterion. The best performing model estimated higher dredge efficiency on fine substrates than on coarse ones (approximately 0.40 and 0.27, respectively). Our results inform the scale of annual abundance estimates from dredge surveys and reduce uncertainty in the sea scallop stock assessments.