A problem with the captive breeding of grouse in this and other studies has been that cocks may attack hens in the confined space of an enclosure. One approach used to avoid severe injury to the hens has been to limit the amount of time each cock spends with his hen. However, separating pairs by hand is labour intensive and disruptive to the grouse. During this study a pen was developed which was separated into two parts by a 60cm tall barrier: this kept the wing-clipped cock on one side but allowed the unclipped hen to flutter to either side and so escape from the cock when she wanted to. The hatchability of eggs produced by captive grouse was unaffected by the type of hatcher used. Both still-air and forced-draught hatchers produced a hatchability of 52-57%. In 1989 and 1990 a total of 107 grouse aged 6-9 weeks was released into a wild grouse population. Data from 22 grouse fitted with radio transmitters showed a similar pattern in survival in both years. Over half died within three weeks of the release date, and most carcasses were discovered in a thin and emaciated condition. After three weeks the main cause of death was predation and only 2 of the 107 released grouse survived to the following breeding season. Initially, released grouse were reluctant to fly, but with time they did take to the wing more readily. In 1989 the release pens were located at the edge of the moor and many released grouse were found on the surrounding farmland. In 1990 the release pens were relocated in the middle of the moor and most grouse were subsequently found on the moor. In 1989 none of the released grouse were shot. In 1990 the total number of grouse shot on the release area was 14 and of this 6 were released grouse. The breeding density of the wild population on the release area declined during the period of the release experiments. However, on an adjacent control area where grouse were not released the breeding density increased. The causes of these patterns were not established.