The prognostic significance of P-wave morphology in patients with coronary artery disease (CAD) is not well-known. A total of 1946 patients with angiographically verified CAD were included in the Innovation to reduce Cardiovascular Complications of Diabetes at the Intersection (ARTEMIS) study. The P-wave morphology could be analyzed in 1797 patients. During 7.4 ± 2.0 years, a total of 168 (9.3%) patients died or experienced resuscitation from sudden cardiac arrest (SCA), 43 (2.4%) patients experienced sudden cardiac death (SCD) or were resuscitated from SCA, 37 (2.1%) patients succumbed to non-SCD (NSCD), and 88 (4.9%) patients to noncardiac death (NCD). Of the P-wave parameters, the absolute P-wave residuum (PWR), the heterogeneity of the P-wave morphology (PWH), and the P-wave duration (Pdur) had the closest univariate association with the risk of SCD/SCA (0.0038 ± 0.0026 vs 0.0022 ± 0.0017, P < .001; 11.0 ± 5.2 vs 8.6 ± 3.6, P < .01; 142.7 ± 16.9 vs 134.8 ± 14.3 milliseconds, P < .01; SCD/SCA vs no SCD/SCA, respectively). After adjustments with factors that were associated with the risk of SCD/SCA, such as diabetes, smoking, left bundle branch block, high-sensitivity C-reactive protein, and high-sensitivity troponin T, PWR (P < .001), PWH (P < .05), and Pdur (P < 0.01) still predicted SCD/SCA but not non-sudden cardiac death. When these parameters were added to the SCD/SCA clinical risk model, the discrimination and reclassification accuracy of the risk model increased significantly (P < .05, P < .001) and the C-index increased from 0.745 to 0.787. The P-wave morphology parameters independently predict SCD/SCA in patients with CAD.