OBJECTIVE: To test whether the association between childhood abuse, cannabis use and psychotic experiences (PEs) was moderated by the COMT (catechol-O-methyltransferase) gene. METHOD: Psychotic experiences (PEs), childhood abuse, cannabis use and COMT Val158Met genotypes were assessed in 533 individuals from the general population. Data were analysed hierarchically by means of multiple linear regression models. RESULTS: Childhood abuse showed a significant main effect on both positive (β = 0.09; SE = 0.04; P = 0.047) and negative PEs (β = 0.11; SE = 0.05; P = 0.038). A significant three-way interaction effect was found among childhood abuse, cannabis use and the COMT gene on positive PEs (β = -0.30; SE = 0.11; P = 0.006). This result suggests that COMT genotypes and cannabis use only influenced PE scores among individuals exposed to childhood abuse. Furthermore, exposure to childhood abuse and cannabis use increased PE scores in Val carriers. However, in individuals exposed to childhood abuse but who did not use cannabis, PEs increased as a function of the Met allele copies of the COMT gene. CONCLUSION: Cannabis use after exposure to childhood abuse may have opposite effects on the risk of PEs, depending on the COMT genotypes providing evidence for a qualitative interaction. Val carriers exposed to childhood abuse are vulnerable to the psychosis-inducing effects of cannabis.