Attempts to control predator numbers through spatially restricted culling typically faces a compensation process via immigration from surrounding source populations. To extend control effort to avoid this issue is in most instances impractical, both logistically and financially. Evidence-based strategy is therefore required to improve management practices. In close collaboration with local managers and hunters, we manipulated culling effort on red fox (Vulpes vulpes) over 5-6 years in 5 areas measuring 246 AE 53 km 2. We estimated fox density in late February each year by spotlight counts with distance sampling and estimated reproductive performance by post-mortem examination of culled foxes. We then used mixed modeling to assess how culling rate (defined as foxes killed/foxes available) affected fox population growth from year to year, accounting for compensatory feedbacks. We found a strong compensatory density feedback acting through immigration, allowing red fox populations to resist high culling rates. Culling appeared ineffective at reducing late winter densities to below 25-32% of the estimated carrying capacity. On average, an annual culling rate equivalent to about 45% of the pre-breeding population was required to maintain density at 1 fox/ km 2 , given a carrying capacity of 1.5 foxes/km 2 , although there was considerable variation among sites. The required culling rate dropped to 25% if the culling could be performed during winter, after the fox dispersal period. In contrast, culling during the pre-dispersal breeding period was totally compensated for through immigration by the following February. Concentrating culling during the winter could improve the ability of practitioners to control year-to-year trends in fox numbers, taking into account site-specific carrying capacity. A winter strategy would also reduce the number of animals killed and hence the ethical and logistical costs of fox control, given limited financial and human resources. Our study illustrates how collaboration between local practitioners and scientists can make large-scale replicated management experiments achievable, leading to mutually approved guidelines.