The prevalence and the clinical meaning of cannabis use in patients with chronic psychosis has not been systematically explored. The authors have compared the diagnostic and symptomatological characteristics of 111 male patients affected by chronic psychosis with and without past or current use of cannabis. Sixty-six patients were still using or had used cannabis; in all cases the use preceded the onset of psychotic symptoms. Forty-three patients were cannabis-positive on urinary screening at the moment of hospitalization and 23 were currently cannabis-free but reported the use of cannabis in the past. Forty-five patients were negative on urinary screening and reported no past history of cannabis use. In evaluating the psychopathological features, the Brief Psychiatric Rating Scale (BPRS) and the Overt Aggression Rating Scale (AORS) were used. The three groups showed similar demographic data, except for age, which was lower in current cannabis users than in nonusers; no differences were found between current and past users. As regards diagnostic features, "mood cluster" was significantly better represented in cannabis users and "schizophrenic cluster" in nonusers; bipolar spectrum disorders were more frequently reported than unipolar ones. When past and current users were grouped together, only blunted affect score was significantly higher in nonusers than in users, while clastic violence showed higher scores in users. These data indicate that chronic, psychosis, whether associated with past or with current use of cannabis, is frequently associated with bipolar spectrum disorders and tends to display less blunted affect and more clastic behavior.