Aim We test whether geographical variation in the length of appendages in rodent species follows predictions of Allen's rule (a positive relationship between appendage length and temperature) at a broad taxonomic scale (order Rodentia). We also test whether the applicability of this rule varies based on the unit of analysis (species or assemblage), the appendage examined (tail, hind foot, ear), body size, occupied habitat, geographical range size, life mode and saltatorial ability. Location World-wide. Time period Current. Major taxa studied Rodents (order Rodentia). Methods We assembled data on the morphology, ecology and phylogeny for ≤ 2,212 rodent species, representing c. 86% of all the described rodent species and c. 95% of the described genera. We tested the predicted Allen's rule associations among size-corrected appendage lengths and both latitudinal and climatic variables (temperature and precipitation). We applied a cross-species approach based on phylogenetic regressions and a cross-assemblage approach based on spatial regressions in equal-area 1.5° grid cells. Results Support for Allen's rule was greatest for the tail and was stronger across assemblages than across species. We detected a negative relationship between tail length and (absolute) latitude, which was accounted for by a positive association between tail length and temperature of the coldest month. This association was greatest in desert species. In addition, we observed a negative relationship between ear length and precipitation. Main conclusions In rodents, Allen's rule is confirmed only for tails, and this association seems to be driven by adaptation to the cold, rather than warm temperatures. Habitat type seems to influence conformity to this rule. Conformity to Allen's rule is likely to be the result of complex evolutionary trade-offs between temperature regulation and other essential species traits.