Background: Despite the rising use of medical cannabis, limited evidence informs the safety and efficacy of cannabis use in cancer patients. A national survey of oncology fellows found that fellows felt insufficiently informed to make clinical recommendations about cannabis. In this secondary analysis of that survey, we aimed to measure how frequently trainees recommend cannabis and determine factors that influence this practice. Given the lack of evidence, we hypothesized that fellows who had been trained in medical cannabis use would be less likely to recommend cannabis to their patients. Methods: An online survey assessing trainee practices on medical cannabis was sent to 155 oncology fellowship program directors between January - March 2021 and directors were asked to distribute it to their fellows. The primary outcome was the number of times oncology fellows recommended cannabis in the prior year. Log-binomial regression was used to calculate risk ratios (RR) and 95% confidence intervals (CI) to evaluate whether the number of cannabis recommendations varied by participant age, gender, legality of cannabis in the state of practice, and prior training in medical cannabis. Results: Nationally, 40 programs from 25 states participated, with 189 of 462 trainees across these programs responding (40% response rate). Twenty-two percent (95% binomial CI: 16.3-29.0%) of participants reported recommending medical cannabis to >5 patients in the past year. Twenty-four percent (95% binomial CI: 18.4-30.5%) of participants had prior training in medical cannabis. Among participant characteristics, only prior training in medical cannabis was associated with recommending cannabis to >5 patients (RR: 2.42; 95% CI: 1.39–4.20). Conclusions: Despite limited evidence to support medical cannabis use among cancer patients, one in five oncology fellows participating in our study recommended it to >5 patients in the past year. Prior training was associated with doubling of rates of recommending medical cannabis in this group. It is thus of critical importance that fellowship training on this subject reflects best available evidence to guide trainees in well-informed decision-making. Factors associated with trainees recommending medical cannabis to >5 patients.

*Adjusted for medical school graduation year, legal status in state of practice, and prior training in medical cannabis.