OBJECTIVE: Studies investigating medication adherence in psychosis are limited by the need of a certain degree of medication adherence and the inclusion of mostly multiple-episode patients. By contrast, noninformed consent, epidemiological studies in first-episode psychosis (FEP) allow the assessment of an important subgroup of patients who persistently refuse antipsychotic medication and thereby never receive an adequate antipsychotic trial. The present study aims to assess the prevalence and predictors of such a "medication refusal" subgroup and its association with illness outcome. METHODS: The present file audit study assessed medication adherence in an epidemiological cohort of 605 FEP patients who were treated within the Early Psychosis Prevention and Intervention Centre for up to 18 months. Medication adherence was categorized into full adherence, nonadherence, and persistent medication refusal. Predictors were analyzed using logistic regression models. RESULTS: During the 18-month treatment period, 204 patients (33.7%) were fully adherent, 287 (47.4%) displayed at least 1 phase of nonadherence, and 114 patients (18.8%) were persistent medication refusers. Poor premorbid functioning, comorbid substance use, and poor insight predicted both medication refusal and nonadherence; a forensic history and no previous contact to psychiatric care were specifically predictive of medication refusal. With respect to illness outcome, nonadherent patients were worse off when compared with fully adherent patients, and medication refusers were even worse off compared with nonadherent patients. CONCLUSIONS: Within a nonselected epidemiological FEP cohort, almost 20% of patients are persistent medication refusers. The found predictors may help to identify the individual risk of persistent medication refusal and may enable an early (preventive) treatment adaptation.