INTRODUCTION: Language impairment has the potential to predict the onset and progression of psychosis. However, it was mainly examined using automated extraction of quantitative linguistic features and their associations with observable psychopathological aspects of psychosis (e.g., formal thought disorders). Little interest has been paid to subjective language disturbances that should phenomenologically anticipate these more objective clinical features. Therefore, the aim of this examination was to investigate subjective language disorders in a Ultra-High Risk (UHR) sample and their associations with clinical and functional outcomes along 2 years of follow-up. METHODS: 170 UHR participants (88 [51.8%] females; mean age = 19.52 ± 6.03 years) were assessed for a broad range of clinical outcomes, including psychosis transition, clinical and functional remission measured with the Positive and Negative Syndrome Scale and the Social and Occupational Functioning Assessment Scale. Comparisons between patients with or without baseline subjective language disorders (specifically explored with the Schizophrenia Proneness Instrument) were analyzed using Chi-Square and Mann-Whitney tests, Kaplan-Meier survival analysis, and binary logistic regression analysis. RESULTS: Across the follow-up, the UHR subgroup with language disorders at entry (n = 80) showed higher and more enduring severity in psychopathology (especially negative and disorganized features), as well as poorer socio-occupational functioning over time. CONCLUSION: The presence of subjective language disturbances at baseline identifies a subgroup of UHR youths with poorer psychopathological and functional prognosis. Further studies examining their association with quantitative linguistic biomarkers are needed, especially to better predict the onset and progression of psychosis.