Understanding the impact of human harvest is fundamental to the effective management of many wildlife populations. Such understanding has been elusive because harvest mortality may compensate for other sources of mortality when the mortality process is density dependent. This problem is exacerbated by the ubiquitous positive correlation between harvest regulations and population size: more harvest is allowed when populations are larger. Connelly et al. (2003) studied the impact of harvest regulations on sage-grouse Centrocercus uropha-sianus using three sets of regulations: closed season, 1-bird bag and seven-day season, 2-bird bag and 23-day season. Connelly et al. (2003) reported a generally negative correlation between harvest regulations and increase in number of males on leks for harvest regulations that ranged from a hunting closure to a 23-day season with a 2-bird bag. Because lek sizes were smaller where hunting was closed there was confounding between harvest and population density, making it difficult to distinguish harvest effects from those of population density. Based on a simple simulation the apparent effects of harvest on change in population size observed by Connelly et al. (2003) could be produced entirely by density-dependent phenomena. Additionally, λ (finite rate of population increase) was greater in areas with more restrictive harvest regulations. λ is a ratio of N_{t 1} to N_t, however, and there is a negative sampling covariance between λ and N_t; we expect λ to be larger when N_t is smaller based purely on this statistical fact. The study by Connelly et al. (2003) is an important attempt to study effects of harvest on population dynamics of sage-grouse. We do not argue that either additive mechanisms in survival or compensatory mechanisms in survival or reproduction influence the relationship between harvest and population dynamics of sage-grouse, but that correlation between population size and harvest regulations, combined with statistical issues make it impossible to distinguish between these two hypotheses in Connelly et al. (2003).