We evaluated survival of elk (Cervus elaphus) calves on 2 contrasting study areas in north-central Idaho, USA, from 1997 to 2004. Recruitment was modest (>30 calves:100 F [calves of either sex:F elk ≥1 yr old]) and stable on the South Fork study area and low (<20 calves:100 F) and declining on the Lochsa study area. The primary proximate cause of calf mortality on both study areas was predation by black bears (Ursus americanus) and mountain lions (Puma concolor). We experimentally manipulated populations of black bears and mountain lions on a portion of each study area. Black bear harvest (harvest density/600km²) initially doubled on the Lochsa treatment after manipulating season bag limits. Mountain lion harvest also increased by 60% but varied widely during the manipulation period. Harvest seasons were closed for black bears and mountain lions on the treatment portion of the South Fork study area. Using the Andersen–Gill formulation (A–G) of the Cox proportional hazards model, we examined effects of landscape structure, predator harvest levels, and biological factors on summer calf survival. We used Akaike's Information Criterion (AIC_c) and multimodel inference to assess some potentially useful predictive factors relative to calf survival. We generated risk ratios for both the best models and for model-averaged coefficients. Our models predicted that calf survival was influenced by biological factors, landscape surrounding calf locations, and predator harvest levels. The model that best explained mortality risk to calves on the Lochsa included black bear harvest (harvest density/600 km²), estimated birth mass of calves, and percentage of shrub cover surrounding calf locations. Incorporating a shrub × time interaction allowed us to correct for nonproportionality and detect that effect of shrub cover was only influential during the first 14 days of a calf's life. Model-averaging indicated that estimated birth mass of calves and black bear harvest were twice as important as the next variables, but age of calves at capture was also influential in calf survival. The model that best explained mortality risk to calves on the South Fork included black bear harvest, age of calves at capture, and gender of calves. Model-averaging indicated that age at capture and black bear harvest were twice as important as the next variable, forest with 33–66% canopy cover (Canopy 33–66). Risk to calves decreased when calves occupied areas with more of this forest cover type. Model-averaging also indicated that increased mountain lion harvest lowered calf mortality risk 4% for every 1-unit increase in lion harvest (harvest density/600 km²) but was lower (<25%) in importance compared to age at capture and black bear harvest. Our results suggest that levels of predator harvest, and presumably predator density, resource limitations expressed through calf birth mass, and habitat structure had substantial effects on calf survival. Our results can be generalized to other areas where managers are dealing with low calf elk recruitment. However, because factors vary spatially, a single management strategy applied in different areas will probably not have the same effect on calf survival.