In a study correlating carnivore density with study area size, Smallwood & Schonewald (1996) suggested that for larger areas, biologists extrapolate from small, high-density study sites to many unstudied low density sites, but that this approach is invalid because of heterogeneous distribution of animals across the landscape. For wolves (Canis lupus), however, size of study areas may also be a consequence of density. Linear regression of data from 18 recent field studies of wolves indicates that study area size is correlated less strongly to mean numbers of wolves counted in an area (r2=0·24) than to wolf density (r2=0·85). When parameters representing biological features of each study area (average wolf pack territory size, ungulate biomass density index and predominant ungulate species), as well as logistical constraints of the study (number of packs), were made available for inclusion through a stepwise procedure, mean numbers of wolves counted were related (r2=0·94) to number of packs, ungulate biomass and deer, while density was related to both study area size and number of wolves counted (r2=0·91). Study area size was found to be strongly predicted (r2=0·96) by ungulate biomass, wolf territory size and wolf numbers. These results clearly reinforce the idea that biological parameters play an essential role in determining animal numbers and density. It also follows, as emphasized by Smallwood & Schonewald (1996, in press), that extrapolation of census or survey information to larger areas needs to be done carefully to account for actual variation in animal density.