The incidental capture, or bycatch, of loggerhead sea turtles Caretta caretta in commercial fishing gear is considered a significant threat to their recovery. Bycatch analyses that use fishery-dependent data only reflect the spatial and temporal co-occurrence of turtles and fishing effort and therefore do not directly reveal conditions associated with turtle distributions. Fisheryindependent and -dependent data can be used together to identify environmental conditions associated with turtle presence and the subsequent risk of a bycatch encounter if fishing effort is present. We developed generalized additive models (GAMs) to describe fishery-independent encounter rates of loggerheads observed in aerial and resource surveys in the US mid-Atlantic region as a function of environmental variables. We then fit a fishery-independent GAM to fishery-dependent data collected from commercial gillnet, bottom trawl, and scallop dredge fisheries in the mid-Atlantic region, and tested the model on new fishery-dependent data to assess how well the model predicted bycatch events. The preferred model describes fishery-independent encounter rates as a function of latitude, sea-surface temperature, depth, and salinity. When this model was fit to fishery-dependent data and tested on new data, it predicted 85% of the observed bycatch events when grouped by latitude and season, although it underestimated bycatch events in southern latitudes in winter. We identify times and areas of elevated bycatch risk on which to focus future conservation efforts and observer coverage.