Recent years have seen increasing scientific interest in whether neuron counts can act as correlates of diverse biological phenomena. Lately, Herculano-Houzel (2022) argued that fossil endocasts and comparative neurological data from extant sauropsids allow to reconstruct telencephalic neuron counts in Mesozoic dinosaurs and pterosaurs, which might act as proxies for behavior and life history traits in these animals. According to this analysis, large theropods such asTyrannosaurus rexwere long-lived, exceptionally intelligent animals with “macaque- or baboon-like cognition” whereas sauropods as well as many ornithischian dinosaurs displayed an ectothermic physiology. Besides challenging established views on Mesozoic dinosaur biology, these claims raise questions on whether neuron count estimates could benefit research on fossil animals in general. Here, we address these findings by revisiting Herculano-Houzel’s (2022) neurological estimates and arguments. We present revised estimates of encephalization and telencephalic neuron counts in dinosaurs, which we derive from phylogenetically informed modeling and an amended dataset of endocranial measurements. For large-bodied theropods in particular, we recover significantly lower neuron counts than previously proposed. Furthermore, we review the suitability of neurological variables such as neuron count estimates and relative brain size to predict cognitive complexity, metabolic rate and life history traits in dinosaurs, coming to the conclusion that they are flawed proxies of these biological phenomena. Instead of relying on such neurological estimates when reconstructing Mesozoic dinosaur biology, we argue that integrative approaches are needed to approach this complex subject.