Context The spatial extent at which landscape variables most strongly influences wildlife populations (i.e., scale of effect) is of key importance to ecology and conservation. Many factors can influence scales of effect, but these relationships are poorly understood. In particular, the way scales of effect vary with different landscape metrics has not been extensively tested. Objectives Our study tested whether metrics of landscape configuration have wider or narrower scales of effect than habitat composition. We also examined how species traits and landscape context influence the magnitude of differences in scales of effect between metrics. Methods We calculated scales of effect for four metrics of habitat configuration and one metric of composition directly from species presence/absence data in online repositories including 102 species of forest mammals from 33 studies across the globe. We then compared the magnitudes and variability of scales of effect using Hedge’s g. Results Two metrics of configuration (flux and patch density) had significantly wider scales of effect than habitat composition, while the differences between the other two metrics of configuration and composition were not statistically significant. The magnitude of the difference between metrics varied among samples, but none of the factors tested explained this variability. Conclusions For forest mammals, the scale of effect of a configuration metric can be expected to be 15–20% wider than a composition metric. However, this pattern may not hold for metrics of configuration that do not quantify attributes of the landscape that mediate species dispersal or that are correlated with the metric of composition.