Lions and other African large carnivores are in decline, due in part to effects of illegal hunting with snares, which can reduce prey availability and directly kill or injure carnivores. It is difficult to effectively remove snares from large ecosystems by patrolling, but an additional approach to reduce effects on large carnivores is to monitor the population closely and de-snare individuals who are found in a snare or have broken free but still carry the wire (often with serious injury). The effectiveness of de-snaring programs to reduce impacts on large carnivores has not been directly tested. Here, we used long-term demographic data from 386 individually identified lions in the Luangwa Valley Ecosystem to test the effects on population growth (λ) and population size (N) of a program to remove snares from injured lions and treat their wounds. Stochastic Leslie matrix projections for a period of five years showed that the population grew with the benefits of de-snaring, but was expected to decline without de-snaring. Mean annual growth (λ) with de-snaring was 1.037 (with growth in 70% of years), closely matching observed changes in population size. Mean annual growth was 0.99 (with growth in 47% of years) for a model that assumed snared animals would have died if not treated, and 0.95 (with growth in 37% of years) for models that also accounted for super-additive effects via the death of dependent cubs and increased infanticide with increased male mortality. De-snaring requires intensive effort, but it can appreciably reduce the effect of snaring on lion population dynamics.