A review of the evidence shows that there is no convincing support for a separate clinical diagnosis of 'cannabis psychosis'. Cannabis can, however, produce brief acute organic reactions and, in moderate to heavy doses, psychotic episodes in clear consciousness. Ingestion in naive users or increasingly heavy use in habitual users can precipitate a schizophreniform episode. Heavy users may have an increased risk of developing schizophrenia in the subsequent 15 years. Well controlled, longitudinal studies are required to explore these associations further and their possible aetiological significance.