We analyzed the leadership behavior of breeding and nonbreeding gray wolves (Canis lupus) in three packs during winter in 1997-1999. Scent-marking, frontal leadership (time and frequency in the lead while traveling), initiation of activity, and nonfrontal leadership were recorded during 499 h of ground-based observations in Yellowstone National Park. All observed scent-marking (N = 158) was done by breeding wolves, primarily dominant individuals. Dominant breeding pairs provided most leadership, consistent with a trend in social mammals for leadership to correlate with dominance. Dominant breeding wolves led traveling packs during 64% of recorded behavior bouts (N = 591) and 71% of observed travel time (N = 64 h). During travel, breeding males and females led packs approximately equally, which probably reflects high parental investment by both breeding male and female wolves. Newly initiated behaviors (N = 104) were prompted almost 3 times more often by dominant breeders (70%) than by nonbreeders (25%). Dominant breeding females initiated pack activities almost 4 times more often than subordinate breeding females (30 vs. 8 times). Although one subordinate breeding female led more often than individual nonbreeders in one pack in one season, more commonly this was not the case. In 12 cases breeding wolves exhibited nonfrontal leadership. Among subordinate wolves, leadership behavior was observed in subordinate breeding females and other individuals just prior to their dispersal from natal packs. Subordinate wolves were more often found leading packs that were large and contained many subordinate adults.