A basic step in understanding the dynamics of a fish population is to quantify movement and mortality rates. Conventional mark-recapture experiments have provided the foundation for studies on animal movement, particularly for fish. Previous studies have shown rapid mixing of sablefish (Anoplopoma fimbria) among fishery regulatory areas, with the pattern of movement related to fish size. Over 300 000 tag releases in Alaska and over 27 000 tag recoveries from 1979 to 2009 were analyzed. We used a Markov model to quantify annual movement probabilities among areas for three size groups of sablefish. The negative-binomial likelihood was used to model the tag-recovery data because of significant overdispersion. Annual movement probabilities were high, ranging from 10% to 88% depending on area of occupancy at each time step and size group. Overall, movement probabilities were very different between areas of occupancy and moderately different between size groups. Estimated annual movement of small sablefish from the central Gulf of Alaska had the reverse pattern of a previous study, with 29% moving westward and 39% moving eastward. Movement probabilities also varied annually, with decreasing movement until the late 1990s and increasing movement until 2009. Year-specific magnitude in movement probability of large fish was highly negatively correlated with female spawning biomass estimates from the federal stock assessment. Mean mortality estimates from time at liberty were similar to the federal stock assessment. Incorporating these tag-recovery and movement data into a fully age-structured spatial stock assessment model will inform harvest apportionment strategies to conserve spawning biomass and maximize future yields.