A recent meta-analysis showed that the mean age of onset of psychosis among cannabis users was almost three years earlier than that of non-cannabis users. However, because cannabis users usually smoke tobacco, the use of tobacco might independently contribute to the earlier onset of psychosis. We aimed to use meta-analysis to compare the extent to which cannabis and tobacco use are each associated with an earlier age at onset of schizophrenia and other psychoses. We also examined other factors that might have contributed to the finding of an earlier age of onset among cannabis users, including the proportion of males in the samples, the diagnostic inclusion criteria and aspects of study quality. The electronic databases MEDLINE, EMBASE, PsycINFO and ISI Web of Science, were searched for English-language peer-reviewed publications that reported age at onset of schizophrenia and other psychoses separately for cannabis users and non-users, or for tobaccosmokers and non-smokers. Meta-analysis showed that the age at onset of psychosis for cannabis users was 32 months earlier than for cannabis non-users (SMD=- 0.399, 95%CI -0.493 - -0.306, z=-8.34, p < 0.001), and was two weeks later in tobacco smokers compared with non-smokers (SMD=0.002, 95%CI -0.094 - 0.097, z=0.03, p=0.974). The main results were not affected by subgroup analyses examining studies of a single sex, the methods for making psychiatric diagnoses and measures of study quality. The results suggest that the association between cannabis use and earlier onset of psychosis is robust and is not the result either of tobacco smoking by cannabis using patients or the other potentially confounding factors we examined. This supports the hypothesis that, in some patients, cannabis use plays a causal role in the development of schizophrenia and raises the possibility of treating schizophrenia with new pharmacological treatments that have an affinity for endo-cannabinoid receptors.