Smith et al. (1999) assessed nest-site selection in Great Horned Owls (Bubo virginianus) by comparing data from 75 owl nests with data recorded at random sites derived from computer-generated map coordinates. The authors concluded that owl nests differed significantly from random sites for nearly all of the eight habitat variables that they measured. I consider some of these conclusions tentative because Smith et al. (1999) ignored the fact that owls do not build their own nests and thus are not free to nest wherever they choose. Although Smith et al.'s approach is adequate for assessing selection of some macrohabitat features, a more appropriate way to assess nest-site selection per se would have been to compare characteristics of owl nests with those measured at a similar number of suitable but unused stick nests in the same study area where the owls nested.