Magnetic Resonance Imaging has advanced our understanding of the neuroanatomical basis of psychosis. However, persistent challenges, including underpowered datasets, recruitment difficulties, high costs, and an overrepresentation of Western populations, hinder progress. Here, we introduce the Psychosis MRI Shared Data Resource (Psy-ShareD), a new open-access repository that aims to address these challenges by harmonising MRI datasets from labs around the world, to build a large globally representative resource. Psy-ShareD provides easily accessible and curated data, allowing a wide variety of analytical approaches. It also includes linked standardised clinical, demographic, and cognitive data, encompassing schizophrenia, bipolar disorder, first-episode psychosis, at-risk cohorts, and controls. Psy-ShareD already hosts over 2000 MRI images; further partner sites are being added. We carefully curate and harmonise the data at the image and feature level to ensure reliability, using HACA3, CycleGAN, and ComBat. We present our validation findings, and results from preliminary analyses, in this poster. Psy-ShareD data will be freely accessible via a straightforward application. We actively encourage access by students and early career researchers; we provide tools and training materials to facilitate analyses. Thus, the Psy-ShareD resource has the potential to significantly increase global research capacity, and to revolutionise our understanding of psychosis through increased power, reliability and reproducibility. It also provides a global perspective, emphasising the importance of cross-ethnic MRI datasets. We encourage researchers with relevant MRI datasets to consider collaborating and contributing their valuable data to the resource; contributors will benefit in terms of authorships. Full information around access/collaboration can be found on: psyshared.com