After roughly a 60-year absence, wolves (Canis lupus) immigrated (1979) and were reintroduced (1995–1996) into the northern Rocky Mountains (NRM), USA, where wolves are protected under the Endangered Species Act. The wolf recovery goal is to restore an equitably distributed metapopulation of ≥30 breeding pairs and 300 wolves in Montana, Idaho, and Wyoming, while minimizing damage to livestock; ultimately, the objective is to establish state-managed conservation programs for wolf populations in NRM. Previously, wolves were eradicated from the NRM because of excessive human killing. We used Andersen–Gill hazard models to assess biological, habitat, and anthropogenic factors contributing to current wolf mortality risk and whether federal protection was adequate to provide acceptably low hazards. We radiocollared 711 wolves in Idaho, Montana, and Wyoming (e.g., NRM region of the United States) from 1982 to 2004 and recorded 363 mortalities. Overall, annual survival rate of wolves in the recovery areas was 0.750 (95% CI  =  0.728–0.772), which is generally considered adequate for wolf population sustainability and thereby allowed the NRM wolf population to increase. Contrary to our prediction, wolf mortality risk was higher in the northwest Montana (NWMT) recovery area, likely due to less abundant public land being secure wolf habitat compared to other recovery areas. In contrast, lower hazards in the Greater Yellowstone Area (GYA) and central Idaho (CID) likely were due to larger core areas that offered stronger wolf protection. We also found that wolves collared for damage management purposes (targeted sample) had substantially lower survival than those collared for monitoring purposes (representative sample) because most mortality was due to human factors (e.g., illegal take, control). This difference in survival underscores the importance of human-caused mortality in this recovering NRM population. Other factors contributing to increased mortality risk were pup and yearling age class, or dispersing status, which was related to younger age cohorts. When we included habitat variables in our analysis, we found that wolves having abundant agricultural and private land as well as livestock in their territory had higher mortality risk. Wolf survival was higher in areas with increased wolf density, implying that secure core habitat, particularly in GYA and CID, is important for wolf protection. We failed to detect changes in wolf hazards according to either gender or season. Maintaining wolves in NWMT will require greater attention to human harvest, conflict resolution, and illegal mortality than in either CID or GYA; however, if human access increases in the future in either of the latter 2 areas hazards to wolves also may increase. Indeed, because overall suitable habitat is more fragmented and the NRM has higher human access than many places where wolves roam freely and are subject to harvest (e.g., Canada and AK), monitoring of wolf vital rates, along with concomitant conservation and management strategies directed at wolves, their habitat, and humans, will be important for ensuring long-term viability of wolves in the region.