In-patients with psychotic symptoms and cannabis-positive urine analysis were assessed by PSE within one week of admission and again at one and six months. Concurrently admitted psychotic patients with drug-free urine analysis were controls. At one week the two groups differed significantly on only five PSE items: changed perception, thought insertion, non-verbal auditory hallucinations, delusions of control, and delusions of grandiose ability. One item (delayed sleep) differed at one month, and none at six months. The symptom cluster at one week is consistent with acute cannabis intoxication. Subjects and controls were mostly single, poorly educated, unemployed people with histories of psychotic disorders, and given major tranquillisers on admission. Compared with controls, subjects were younger, less likely to have psychiatric histories, more often male, Afro-Caribbeans with a history of convictions and compulsory admissions. The commonest diagnosis was schizophrenia. Use of the label 'cannabis-induced psychosis' may obscure a diagnosis of paranoid schizophrenia. A short-lived psychotic episode does occur in clear consciousness after cannabis intoxication, but chronic cannabis-induced psychosis was not found.