To characterize cannabis use among cancer patients, we aimed to describe 1) patterns of cannabis use across multiple cancer sites; 2) perceived goals, benefits, harms of cannabis; and 3) communication about cannabis. Patients with 9 different cancers treated at Memorial Sloan Kettering Cancer Center between March and August 2021 completed an online or phone survey eliciting cannabis use, attitudes, and communication about cannabis. Multivariable logistic regression estimated the association of cancer type and cannabis use, adjusting for sociodemographic characteristics and prior cannabis use. Among 1258 respondents, 31% used cannabis after diagnosis, ranging from 25% for lung cancer to 59% for testicular cancer. Characteristics associated with cannabis use included younger age, lower education level, and cancer type. In multivariable analysis, compared to lung cancer patients, gastrointestinal cancer patients were more likely to use cannabis (odds ratio [OR] 2.64, 95% confidence interval [CI] 1.25-5.43). Cannabis use in the year prior to diagnosis was strongly associated with cannabis use after diagnosis (OR 19.13, 95% CI 11.92-30.72). Among users, reasons for use included difficulty sleeping (48%); stress, anxiety, or depression (46%); and pain (42%). Among respondents who used cannabis to improve symptoms, 70-90% reported improvement; < 5% reported that any symptom worsened. Only 25% discussed cannabis with healthcare providers. Almost a third of cancer patients use cannabis, largely for symptom management. Oncologists may not know about their patients' cannabis use. To improve decision making about cannabis use during cancer care, research is needed to determine benefits and harms of cannabis use.