A small number of studies have investigated predictive factors in relation to the Global Assessment of Functioning (GAF) scale. This study aimed to explore the influence of clinical and socio-demographic factors in a psychiatric inpatient setting in relation to treatment outcome measured by the GAF. The studied psychiatric inpatient sample consisted of 816 episodes of care, with GAF ratings made at both admission and discharge. Multiple linear regressions were performed to analyze what variables predicted GAF scores at admission and at discharge. Significant predictors of GAF scores at admission were age, schizophrenia, other psychotic disorders, and no registered diagnosis. GAF scores at admission, patients’ diagnoses, and ward affiliation were able to significantly predict GAF at discharge. Specialized wards did not necessarily deliver the best treatment results in spite of their diagnostic specialization. This study provides support to the construct validity of the GAF when used as a measure of outcome.