Pre-clinical research supports that cannabinoids reduce opioid dose requirements, but few studies have tested this in humans. This review evaluates ecological and epidemiological studies that have been cited as evidence that medical cannabis use may reduce opioid use and opioid-related harms. Medline and Embase were searched for relevant articles. Data were extracted on study setting, analyses approach, covariates, and outcomes. Eleven ecological and 14 epidemiological studies were found. In ecological studies, states that allow medical cannabis laws have reported a slower rate of increase in opioid overdose deaths compared with states without such laws. These differences have increased over time and persisted after controlling for state sociodemographic characteristics and use of prescription monitoring programmes. Few studies have controlled for other potential confounders such as opioid dependence treatment and imprisonment rates. Some epidemiological studies provide evidence that cannabis availability may reduce opioid use, but are limited by selection bias, cross-sectional designs, and self-reported assessments of the opioid-sparing effects of cannabis. Some epidemiological and ecological studies suggest that cannabis may reduce opioid use and harms, although important methodological weaknesses were identified. Well-designed clinical studies may provide more conclusive evidence on whether cannabinoids can reduce opioid use and related harm.