The hominoid scaphoid is an integral part of the wrist bridging the proximal and distal carpal rows and helps maintain stability of the wrist joint alongside dissipating loads from the thumb during both locomotory and manipulative behaviours. Variation in scaphoid morphology, including facet shape and orientation, is related to its functional role within the wrist and hand use overall, and has played a key role in hypotheses about the origin of human bipedalism and evolution of tool use. However, scaphoid morphology is irregular and complex, making it difficult to capture intra-and interspecific shape variation from linear measurements alone, potentially missing functionally-informative morphology. This study quantifies the overall morphology of the scaphoid using 3D geometric morphometrics (GM), with anatomical and sliding semi-landmarks, to assess shape differences across African apes, humans and fossil hominins. Surface models were generated from microcomputed tomography scans in a sample of Pan troglodytes (N=7), Pan paniscus (N=7), Gorilla gorilla (N=14), and Homo sapiens (N=20). Within this comparative context, we investigated the potential functional signals in scaphoid morphology in Australopithecus sediba (MH2), Homo floresiensis (LB1), Australopithecus sp. (StW 618), Homo naledi (UW101-1726, UW 102-117R) and Homo neanderthalensis (Kebara 2, El Sidron SD1243 & SD258, Tabun C1). Due to slightly poorer preservation of theLB1, UW101-1726 and Tabun C1 scaphoids, multiple reconstructions of missing morphology were created in Geomagic Wrap 2021, to allow for comparisons of the whole scaphoid rather than selected articular surfaces. Landmarking was done manually in Avizo 6.3 and the GM and statistical analyses were run in R using the package Morpho. Principal components analysis (PCA) using Procrustes coordinates was carried out in shape space, to quantify shape differences across species. Results of a permutational ANOVA of the extant sample reveal significant differences in shape for all pairwise comparisons, excluding bonobos and chimpanzees. The PCA yields clear separation among extant taxa; the first PC separates African apes from humans, largely reflecting the expanded trapezium-trapezoid facet in the latter, while the second PC separates Panfrom Gorilla, due to a larger tubercle and more circular radial and capitate facet in the latter. When looking at the morphospace of PC1 and PC2, StW 618 falls within the Gorilladistribution, reflecting a more palmarly projecting tubercle (creating a deeper carpal arch) and less-expanded trapezium facet relative to other fossil hominins and humans. Neandertals cluster close to extant humans, but the El Sidron specimens in particular, aredistinguished by a small lunate facet. The MH2, LB1 and H. naledispecimens fall out as intermediate between the African ape and human distributions on PC1 and are separated from all other taxa on PC3, but each reflects a slightly different suite of scaphoid features. LB1 has a longer tubercle with a more concave trapezoid facet. A. sediba and H. naledi share greater similarity in scaphoid shape, despite temporal differences, with the trapezium facet expanding onto the tubercle and lunate facet shape but diverge in tubercle orientation and robusticity. An expanded trapezium facet allows for better force distribution on the radial side when the thumb is loaded in an abducted position. This coupled with a less palmarly-oriented tubercle suggests a scaphoid morphology better adapted for tool production. Each of the fossil scaphoids examined show different combinations of joint shape and orientation indicating the evolution of hominin wrist occurred in a mosaic manner in response to potentially different locomotory and manipulative capabilities.