We developed models and provide computer code to make carcass recovery data more useful to wildlife managers. With these tools, wildlife managers can understand the spatial, temporal (e.g., across time periods, seasons), and demographic patterns in mortality causes from carcass recovery datasets. From datasets of radio-collared and non-collared carcasses, managers can calculate the detection bias by mortality cause in a non-collared carcass dataset compared to a collared carcass dataset. As a first step, we provide a standard procedure to assign mortality causes to carcasses. We provide an example of these methods for radio-collared wolves (n = 208) and non-collared wolves (n = 668) found dead in Wisconsin (1979–2012). We analyzed differences in mortality cause relative to season, age and sex classes, wolf harvest zones, and recovery phase (1979–1995: initial recovery, 1996–2002: early growth, 2003–2012: late growth). Seasonally, illegal kills and natural deaths were proportionally higher in winter (Oct–Mar) than summer (Apr–Sep) for collared wolves, whereas vehicle strikes and legal kills were higher in summer than winter. Spatially, more illegally killed collared wolves occurred in eastern wolf harvest zones where wolves reestablished more slowly and in the central forest region where optimal habitat is isolated by agriculture. Natural mortalities of collared wolves (e.g., disease, intraspecific strife, or starvation) were highest in western wolf harvest zones where wolves established earlier and existed at higher densities. Calculating detection bias in the non-collared dataset revealed that more than half of the non-collared carcasses on the landscape are not found. The lowest detection probabilities for non-collared carcasses (0.113–0.176) occurred in winter for natural, illegal, and unknown mortality causes.