Cross‐boundary management issues concerning the spatial requirements of large predators and the inadequate size of protected areas are exemplified by the movements of gray wolf (Canis lupus) in and out of Algonquin Park, Canada. From 1987 to 1993, 65 juvenile and adult wolves from 22 packs inside Algonquin were radio‐tagged and regularly monitored by aerial and ground telemetry. Of the 41 wolves that died during the 6‐year study, mortality caused by humans was the highest single source, at 56%. Most mortality (68%) occurred outside the park, of which almost half (48%) was related to the seasonal movements by wolves out of the park in response to migrating white‐tailed deer (Odocoileus virginianus). In 1993 the provincial government banned the killing of wolves in a 200‐km2 deer wintering area adjacent to the park. The ban was considered necessary because (1) a high proportion (82%) of packs from the eastern half of the park was traveling to the deer area; (2) outside the park killing of wolves by hunting or trapping was uncontrolled; and (3) much of the mortality comprised adult, non‐dispersing wolves. Public response to the ban was hostile initially, but apart from some flagrant violations the protective measures appear to have been successful. Only 2 radio‐collared wolves were killed in 1993 and 1994 combined, compared with 10 in 1992 alone. Controversy over management decisions led us to conclude that cross‐boundary problems associated with this controversial species necessitate complete, partial, or emergency protection. Complete protection requires protection beyond the existing park boundary, either through buffer zones or increased regulation. Partial and emergency protection strategies rely on effective monitoring, an important factor if public cooperation is limited. At Algonquin the magnitude of wolf killing, the importance of the wolf in interpretive programs, and the difficulties in regulating partial protection because of unreported kills and damage to radio collars makes complete protection the prudent choice.