Abstract Background Epidemiological studies show a dose–response association between cannabis use and the risk of psychosis. This review aimed to determine whether there are identifiable risk-thresholds between the frequency of cannabis use and psychosis development. Methods Systematic search of Embase, MEDLINE, PsycINFO, CINAHL, and Web of Science for relevant studies (1 January 2010–26 April 2021). Case–control or cohort studies that investigated the relationship between cannabis use and the risk of psychosis development that reported effect estimates [odds ratios (OR), hazard ratios (HR), risk ratios (RR)] or the raw data to calculate them, with information on the frequency of cannabis consumption were included. Effect estimates were extracted from individual studies and converted to RR. Two-stage dose–response multivariable meta-analytic models were utilized and sensitivity analyses conducted. The Newcastle Ottawa Scale was used to assess the risk of bias of included studies. Results Ten original (three cohorts, seven case–control) studies were included, including 7390 participants with an age range of 12–65 years. Random-effect model meta-analyses showed a significant log-linear dose–response association between cannabis use frequency and psychosis development. A restricted cubic-splines model provided the best fit for the data, with the risk of psychosis significantly increasing for weekly or more frequent cannabis use [RR = 1.01, 95% confidence interval (CI) 0.93–1.11 yearly; RR = 1.10, 95% CI 0.97–1.25 monthly; RR = 1.35, 95% CI 1.19–1.52 weekly; RR = 1.76, 95% CI 1.47–2.12 daily] Conclusion Individuals using cannabis frequently are at increased risk of psychosis, with no significant risk associated with less frequent use. Public health prevention messages should convey these risk-thresholds, which should be refined through further work.