Functional impairment is a defining feature of psychotic disorders. The Functional Assessment Short Test (FAST) is one of the most widely used instruments to measure psychosocial functioning. However, cut-offs of impairment have been well-established for bipolar disorders, but not for other clinical populations. This study aims to analyse psychometric properties of the FAST and establish their corresponding cut-off values for the different severity gradations in a first-episode of non-affective psychosis (FEP) patients. Global Assessment Functioning (GAF) and FAST ratings from 212 non-affective FEP and 204 healthy controls were analyzed. The psychometric properties of FAST (internal consistency, concurrent validity, discriminant validity, factorial analyses and sensitivity to change) were analyzed. The severity gradations of the FAST were defined by the congruence between two grading methods: linear regression analysis (LRA) and percentiles. The FAST showed strong psychometric properties. LRA with the GAF scores yielded the following equation: GAF= 80.83 - 0.639*FAST. The FAST ranges in non-affective FEP patients derived from LRA and percentiles, were as follows: 0-9 (No impairment); 10-19 (Minimal impairment); 20-34 (Mild impairment); 35-45 (Moderate impairment); 46-72 (Severe impairment). Patients with no functional impairment had a higher socioeconomic status, fewer depressive and negative symptoms, lower severity of illness and higher cognitive reserve level than the others groups. In conclusion, the FAST shows optimal psychometric properties which corroborate its applicability in FEP populations. It is a well-demonstrated valid instrument and the present cut-off scores could be implemented in clinical and research practice to assess properly the psychosocial functional outcome of non-affective FEP populations.