The PANSS Autism Severity Score (PAUSS) is a popular measure of autistic features in First Episode Psychosis (FEP) samples. However, evidence on its longitudinal stability, course and treatment response is poor. Therefore, the main aim of this research was to compare clinical outcomes between FEP individuals with or without “autistic features” enrolled within an “Early Intervention in Psychosis” (EIP) service across 2 years of follow-up, as well as any significant association with EIP treatment components. FEP subjects completed the Positive And Negative Syndrome Scale (PANSS), the Global Assessment of Functioning (GAF), and the Health of the Nation Outcome Scale (HoNOS) at entry and across the follow-up. Statistical tests included Kaplan-Meyer survival analysis, mixed-design ANOVA, and multiple linear logistic regression analysis. 301 FEP subjects were enrolled (85 [28.0%] scored above the PAUSS cut-off score). Across the follow-up, the PAUSS + subgroup showed lower incidence rates of both symptomatic and functional remission. No PAUSS long-term stability was observed, but a statistically significant reduction in its values. This longitudinal change was mainly predicted by the total number of case management sessions offered within the EIP program. Our results suggest that the PAUSS could not represent a valid instrument to assess “trait-like” autistic features in FEP subjects. On contrary, it seems to capture a FEP subgroup characterized by higher severity levels in psychopathology and poorer outcomes and prognosis.