Background. There is evidence that cannabis use might be relevant to the aetiology of schizophrenia. We aimed to measure any change in cannabis use over time in those first presenting with schizophrenia in South-East London from 1965 to 1999, and compare this with change in use in those presenting with non-psychotic psychiatric disorders.

Method. The rate of cannabis use in the year prior to first ever presentation was measured over seven time periods. Logistic regression modelling was used to determine (a) whether cannabis use changed over time, after controlling for age, sex and ethnicity, and (b) whether there was an interaction between diagnosis and time.

Results. Cannabis use increased over time in both the schizophrenia group [odds ratio per time period (OR) 2·03, 95% confidence interval (CI) 1·74–2·38, p<0·0001] and the non-psychotic disorders group (OR 1·24, 95% CI 1·05–1·47, p=0·012), after controlling for age, sex and ethnicity. However, the effect of time was significantly greater in the schizophrenia group than in the non-schizophrenia group (χ2=17, p<0·0001).

Conclusion. Cannabis use in the year prior to presentation with schizophrenia increased markedly between 1965 and 1999, and disproportionately so compared to increase in cannabis use in other psychiatric disorders.