Securing the long-term acceptance of large carnivores such as the wolf (Canis lupus) in Europe and North America raises a difficult challenge to conservation biologists: planning removals to reduce depredations on livestock while ensuring population viability. We use stochastic-stage-structured population models to investigate wolf population dynamics and to assess alternative management strategies. Among the various management strategies advocated by agencies, zoning that involves eliminating wolves outside a restricted area should be designed with caution, because probabilities of extinction are extremely sensitive to the maximum number of packs that a zone can support and to slight changes in stage specific survival probabilities. In a zoned population, viability is enhanced more by decreasing mortality rates in all classes than by increasing wolf zone size. An alternative to zoning is adaptive management, where there is no limit on pack number but population control can be operated whenever some predefined demographic conditions are met. It turns out that an adaptive management strategy that removes a moderate percentage (10%) of the population following each year of more than 5% of total population growth would provide visible actions addressing public concerns while keeping extinction probability low.