Primary tropical forests are lost at an alarming rate, and much of the remaining forest is being degraded by selective logging. Yet, the impacts of logging on biodiversity remain poorly understood, in part due to the seemingly conflicting findings of case studies: about as many studies have reported increases in biodiversity after selective logging as have reported decreases. Consequently, meta-analytical studies that treat selective logging as a uniform land use tend to conclude that logging has negligible effects on biodiversity. However, selectively logged forests might not all be the same. Through a pantropical meta-analysis and using an information-theoretic approach, we compared and tested alternative hypotheses for key predictors of the richness of tropical forest fauna in logged forest. We found that the species richness of invertebrates, amphibians, and mammals decreases as logging intensity increases and that this effect varies with taxonomic group and continental location.