AbstractFishing mortality and natural mortality rates for Atlantic HalibutHippoglossus hippoglossuson the Scotian Shelf and southern Grand Banks were estimated from a multiyear tagging study. Models that were used to estimate mortality rates incorporated tag loss. Between 2006 and 2008, 1,913 Atlantic Halibut were double‐tagged with t‐bar anchor tags; as of 26 August 2010, 368 of these fish had been recaptured. We estimated instantaneous fishing mortality (F) separately for each cohort in the first year (on average, 6 months) after release to allow newly tagged animals to mix with the population. A two‐parameter model was used to describe tag loss. Tag loss was estimated at 13% per year in the first year and 10% per year in the second and subsequent years. Using the multiyear model with incomplete mixing and assuming 90% tag reporting and 80% survival from tagging, average instantaneous natural mortality (M) of Atlantic Halibut was estimated to be 0.22 andFwas estimated to be 0.15 in 2007, 0.24 in 2008, and 0.18 in 2009. These estimates ofFwere comparable to those from the stock assessment population model. However, the estimates ofMwere higher than inferred estimates ofMbased on life history and growth. Estimates ofFandMwere sensitive to the minimum size of Atlantic Halibut at the time of release. An increase inFwith size is consistent with fishery size selectivity resulting from either gear selectivity or the distribution of fishing effort where there is spatial heterogeneity in the size composition. We suggest thatMmay have been overestimated because of emigration from the study area.