Recovery rates were determined for marked fish released into 10 of 15 cove areas and 2 of 6 open water areas during the cove rotenone study at Crooked Creek Bay, an 85 ha arm of Barkley Lake, Kentucky. Fish were tagged with a F10y FD-68B dart tag and either released back into the same study area or introduced from an outside area. Of 986 fish tagged in the embayment, 89% were recaptured. Seventy-five percent of the marked fish were recovered in 9 cove areas, where marked fish were released back into the same cove section. Only 32% were recovered from a back-cove area in which marked fish were introduced from an open water area. Sixty percent were recovered outside this area, indicating a strong tendency for displaced fish to escape. Percentage escapement of marked fish from the other 9 cove areas was only 16%. Escapement increased in each cove section that was progressively farther from open water. Escapement also decreased as mean depth of coves increased. This relationship, when expressed in a multiple regression, was significant (P>O.04). The marking and release offish taken from within a study cove is recommended versus displacement to achieve a higher percent recovery. Selection of cove areas having a mean depth resembling that of adjacent open water habitat is also suggested to minimize escapement and obtain a more representative sample of the reservoir fish population.