In the present study we sought: 1) to estimate the frequency of substance use disorders (SUD), and 2) to investigate whether there is a mere association between diagnosis and SUD in a large cohort of patients with severe psychiatric disorders representative of the usual setting and modality of care of a psychiatric emergency service in a geographically well-defined catchment area in Italy, independent of sociodemographic features, anamnestic data and clinical status. The study was conducted between January 2007 and December 2008. The following rating scales were performed: the Clinical Global Impression-Severity (CGI-S), the Global Assessment of Functioning scale (GAF) and the Brief Psychiatric Rating Scale (BPRS). Factors found to be associated (p<0.05) with SUD[+] in the univariate analyses were subjected to multilevel logistic regression model with a backward stepwise procedure. Among 848 inpatients of our sample 29.1% had a SUD codiagnosis. Eleven factors accounted for 30.6% of the variability in SUD[+]: [a] a Personality Disorder diagnosis, [b] a Depressive Disorder diagnosis, [c] male gender, [d] previous outpatient contacts, [e] single marital status, [f] no previous psychiatric treatments, [g] younger age, [h] lower scores for BPRS Anxiety-depression and [i] BPRS Thought Disturbance, [l] higher scores for BPRS Activation and [m] BPRS Hostile-suspiciousness. The findings are important in identifying (1) the complexity of the clinical presentation of SUD in a inpatients sample, (2) the need for collaboration among health care workers, and (3) the need to develop and apply treatment programs that are targeted at particular risk groups., (Copyright © 2014 Elsevier Inc. All rights reserved.)