BACKGROUND AND HYPOTHESIS: Animal, post-mortem, and pharmacological studies suggest altered choline levels in psychosis. Proton magnetic resonance spectroscopy (1H-MRS) may allow in vivo investigation of the cholinergic system. This meta-analysis examined the magnitude and variability of 1H-MRS choline across the psychosis spectrum. STUDY DESIGN: Following a pre-registered protocol (CRD42023403879), we searched MEDLINE, EMBASE, and PsycINFO for studies reporting 1H-MRS choline levels in individuals with psychosis, clinical high risk (CHR) states, and healthy controls. Standardized mean differences (SMDs) and log coefficient of variation ratios were pooled using random-effects meta-analysis, stratified by brain region and clinical subgroup. Meta-regression examined associations with imaging and clinical parameters. STUDY RESULTS: We included 165 studies comprising 5178 patients and 4269 controls. Choline levels were elevated in people with psychosis in the anterior cingulate cortex (ACC: SMD = 0.23, CI 0.12-0.33), medial prefrontal cortex (mPFC: SMD = 0.12, CI 0.01-0.23), and dorsolateral prefrontal cortex (dlPFC: SMD = 0.15, CI 0.01-0.28). Elevations were also seen in CHR cohorts (ACC: SMD 0.19, CI 0.03-0.36; mPFC: SMD 0.36, CI 0.16-0.57), with the largest effects in treatment-resistant schizophrenia (ACC: SMD 0.65, CI 0.38-0.91; dlPFC: SMD 1.13, CI 0.65-1.60). Increased choline variability was observed in the dlPFC in psychosis cohorts and the mPFC and temporal lobe in CHR groups. CONCLUSIONS: This is the largest meta-analysis of 1H-MRS choline in psychosis to date. Increased variability in psychosis and greater differences in treatment-resistant cohorts suggest increased choline levels might identify a subgroup who do not respond to dopamine antagonist treatment.