Patients with non-valvular atrial fibrillation (NVAF) and chronic kidney disease (CKD) are at increased risk of stroke and bleeding. Although direct oral anticoagulant (DOAC) trials excluded patients with severe CKD, a growing portion of CKD patients have been starting DOACs and limited data from real-world outcome in this high-risk setting are available. The INSigHT registry included 632 consecutive NVAF patients that started apixaban (256 patients, 41%), dabigatran (245, 39%) and rivaroxaban (131, 20%) between 2012 and 2015. Based on creatinine clearance, two sub-cohorts were defined: (1) non-CKD group (CrCl 60–89 mL/min, 413 patients) and (2) CKD group (15–59 ml/min, 219). Compared to non-CKD patients, those with CKD, were at higher ischemic (CHA₂DS₂-VASc 4.5 vs 2.9, p < 0.001) and hemorrhagic risk (HAS-BLED 2.4 vs 1.8, p < 0.001). At 2-year follow-up, the overall ISTH-major bleeding and thromboembolic event rates were 5.2% and 2.3% and no significant difference between non-CKD and CKD patients for both efficacy and safety endpoints were observed. In non-CKD patients, the 2-year ISTH-major bleeding rates were higher in rivaroxaban group (HR 2.9, 95% CI 1.1–7.3; p = 0.047) while dabigatran showed non-significant excess in thromboembolic events (HR 4.3, 95% CI 0.9–20.8; p = 0.068). In CKD patients, a significantly higher rate of thromboembolic events was observed in rivaroxaban (HR 6.3, 95% CI 1.1–38.1; p = 0.044). This real-world, non-insurance database registry shows remarkable 2-year safety and efficacy profile of DOACs even in patients with moderate to severe CKD. Head to head differences between DOACs are exploratory, hypothesis generating and warrant further investigation in larger studies.