Background Few data are available on direct oral anticoagulant (DOAC) use in patients with cancer and atrial fibrillation (AF). Methods We retrospectively analyzed prospectively collected data from a single-center registryon 2,272 patients who tookDOACs for AF (apixaban:1,014; edoxaban:267; rivaroxaban:498; dabigatran:493). Patients were monitored for 2 years andclassified into non-cancer (n=2009) and cancer group (n=263) (cancer onset during DOAC treatment, active canceratDOAC administration, and cancer history).Major bleeding (MB) and thromboembolic events (TEEs) were evaluated. Results The mean age was 73±10 years. CHADS2 and HAS-BLED scores were 1.95±1.32 and 1.89±0.96,respectively.In the present study, the prevalence of gastrointestinal and genitourinary cancer was 61% and 8%, respectively.The MB and TEEs incidences were 2.4 and 2.2 per 100-patient years, respectively. The appropriate dosing rate, body weight, and Ccrvalue in cancer patients were significantly lower than those in non-cancer patients. Cancer patients were significantly older than non-cancer patients. In MB patients diagnosed with gastrointestinal or genitourinary cancer during follow-up, the clinically relevant bleeding such as melena or hematuria occurred.Additionally, there was a significantly higher MB incidence in cancer patients than in non-cancer patients (p<0.01). Conclusions AF patients with cancer was associated with a higher risk of MB compared with those without cancer despite higher rate of inappropriate low dose. Bleeding such as melena and hematuria after DOAC administration might suggest that the symptoms are associated with cancer of the site.