Ptarmigan and grouse species (Lagopus spp.) are thought to be able to compensate for a modest harvest because there is a surplus of breeding birds that are prevented from breeding by territory holders. To estimate the degree of harvest-mortality compensation reliably we experimentally harvested 0%, 15% and 30% of the willow ptarmigan (Lagopus lagopus) on 13 estates ranging from 20 to 54 km2 in size during four hunting seasons in Norway according to a regional block design. Population overwinter growth rate was strongly negatively density dependent, but despite this, and contrary to earlier findings, only 33% of the harvest was compensated for. The lack of compensation was probably caused by long-distance juvenile dispersal that was unaffected by the harvest. The need for large-scale management experiments to detect the effects of harvest was clearly demonstrated: lack of compensation was found only when we used the whole dataset and not when the data were analysed by year or block. Our study shows that it is very difficult to demonstrate a population's lack of harvest compensation and warns against using small-scale, out-of-season or poorly replicated studies as a basis for future harvest-management decisions.