Ecological theory and wildlife management often depend on reliable com
parison and interpretation of population density estimates. A synthesi
s of 1,772 mammalian carnivore population estimates (713 unique to ref
erence, species, site, and size of study area) from 74 species reveale
d global patterns among aspects of study and interpretive design that
undermine the reliability and usefulness of density comparisons. The s
patial extent of the study area could explain most of the variation in
density, probably because study areas are typically delineated around
population clusters. We related the scale-defined density estimates (
regression residuals) to 28 other variables measured from the publishe
d literature, but none provided convincing biological explanation of t
he variation in density. Many aspects of study and interpretive design
were possibly ill-suited to identifying the factor(s) influencing den
sity. Study attributes and findings were reported inconsistently. and
were subject to ideological motivations. Descriptions of vegetation we
re most difficult to relate to density. More intensive sampling and es
timation methods produced above-average density estimates, but the dif
ferences were slight and the evidence lacking for concluding whether t
hese more intensive methods were also more accurate. The first underly
ing factor extracted from principle-components analysis described the
growing recognition of population declines and range reductions among
large-bodied carnivores, which has also influenced study design. Anoth
er factor described an increasing trend for density to be compared and
extrapolated to larger areas, but without adjusting for the effect of
scale. To understand the factors influencing carnivore distribution a
nd abundance, sampling and reporting methods (e.g., site description w
ith maps) will need to represent the available pool of species, locati
ons, and ecological conditions at larger-than-conventional spatial and
temporal scales.