Aim: To examine possible causal linkages between cannabis use and psychosis using data gathered over the course of a 25-year longitudinal study.

Design: A 25-year longitudinal study of the health, development and adjustment of a birth cohort of 1265 New Zealand children (635 males, 630 females).

Setting: The Christchurch Health and Development Study, a general community sample.

Participants: A total of 1055 participants from the Christchurch Health and Development Study (CHDS) cohort for whom data on cannabis use and psychotic symptoms were available on at least one occasion from 18, 21 and 25 years.

Measurements: As part of this study, data were gathered on frequency of cannabis use and psychotic symptoms at ages 18, 21 and 25 years.

Findings: Regression models adjusting for observed and non-observed confounding suggested that daily users of cannabis had rates of psychotic symptoms that were between 1.6 and 1.8 times higher (P < 0.001) than non-users of cannabis. Structural equation modelling suggested that these associations reflected the effects of cannabis use on symptom levels rather than the effects of symptom levels on cannabis use.

Conclusions: The results of the present study add to a growing body of evidence suggesting that regular cannabis use may increase risks of psychosis. The present study suggests that: (a) the association between cannabis use and psychotic symptoms is unlikely to be due to confounding factors; and (b) the direction of causality is from cannabis use to psychotic symptoms.