Schizophrenia is a debilitating psychiatric disorder presenting with delusions, hallucinations, and cognitive dysfunction, characterised by widespread aberrant connectivity both in early and chronic stages. One of the relatively most connected forebrain regions is the claustrum, which has been shown to have reduced volume in schizophrenia. However, whether claustrum microstructure and connectivity are altered remains unknown. We hypothesized aberrant claustrum structure and tract-based connectivity in individuals with recent-onset schizophrenia. Structural and diffusion-weighted MRI data from 48 controls and 48 participants with non-affective early psychosis (NA-EP) from the Human Connectome Project, diagnosed with schizophrenia or related disorders within the last five years, were preprocessed and analysed using FSL. The claustrum was automatically segmented using a deep learning model. Individual fractional anisotropy (FA) maps were computed and probabilistic tractography was performed to derive measures of connection probability. Language function and episodic memory had been assessed using the NIH Toolbox Picture Vocabulary Test and Picture Sequence Memory Test, respectively.We found significantly reduced bilateral claustrum volumes relative to total intracranial volume as well as lower mean FA of the left claustrum in NA-EP subjects compared to healthy controls. Furthermore, connection probability was reduced between both claustra and ipsilateral temporal association cortices. Finally, aberrant claustrum structure and connectivity correlated with impaired language abilities and episodic memory scores.These results demonstrate structural alterations and dysconnectivity of the claustrum, which are associated with language and memory impairments, in the early stages of schizophrenia.