Background/Question/Methods The latitudinal gradient in species diversity (LGSD) has long been known but lacks a general, mechanistic explanation. Assuming that generality of pattern may reflect generality of process, we examine LGSDs of New World mammal orders and families, testing for generality of pattern in two components of the LGSD, shape and peak. Using the overall mammal pattern as the expectation, subtaxa should exhibit a tropical peak and a concave-down pattern of diversity. Results/Conclusions Three of 12 mammal orders and 23 of 48 families examined here failed to show both the peak and shape expected. Thus the overall LGSD is not a general pattern across mammal subtaxa and therefore it is unlikely that the LGSD is generated by a shared general mechanism or historical process. Bats and taxa that originated in South America before the emergence of the Isthmus of Panama are more likely to diversity peaks within the tropics, arguing that the idiosyncratic biogeographical history of taxa may be more important to generating the LGSD than macroecological mechanistic explanations.