Body size is of fundamental importance to our understanding of extinct organisms. Physiology, ecology and life history are all strongly influenced by body size and shape, which ultimately determine how a species interacts with its environment. Reconstruction of body size and form in extinct animals provides insight into the dynamics underlying community composition and faunal turnover in past ecosystems and broad macroevolutionary trends. Many extinct animals are known only from incomplete remains, necessitating the use of anatomical proxies to reconstruct body size and form. Numerous limitations affecting the appropriateness of these proxies are often overlooked, leading to controversy and downstream inaccuracies in studies for which reconstructions represent key input data. In this perspective, we discuss four prominent case studies (Dunkleosteus, Helicoprion, Megalodon and Perucetus) in which proxy taxa have been used to estimate body size and shape from fragmentary remains. We synthesise the results of these and other studies to discuss nuances affecting the validity of taxon selection when reconstructing extinct organisms, as well as mitigation measures that can ensure the selection of the most appropriate proxy. We argue that these precautionary measures are necessary to maximise the robustness of reconstructions in extinct taxa for better evolutionary and ecological inferences.