Background: Diabetes mellitus (DM) is an independent risk factor for atrial fibrillation (AF), the most frequent clinical arrhythmia, but the mechanism(s) underlying this clinical association are unknown. Increased reactive oxygen species (ROS) and O-GlcNAcylation (OGN) is a hallmark of DM. Hyperactivated CaMKII promotes AF, and CaMKII may become constitutively active and proarrhythmic by oxidation (ox-CaMKII) or OGN (OGN-CaMKII), post translational modifications to adjacent amino acids in the regulatory domain. We hypothesize that CaMKII promotes diabetes-induced AF through ox-CaMKII and OGN-CaMKII. Methods and Results: We generated knock-in mouse models, based on the predominant CaMKII isoform in myocardium (CaMKIIδ), to selectively eliminate ox-CaMKII and OGN-CaMKII. We used the pancreatic β-cell toxin, streptozocin (STZ) or a combination of high fat diet and STZ, to generate validated mouse models of T1D and T2D, and an established right atrial pacing model of paroxysmal AF. T1D and T2D mice showed increased AF [70% (14/20), p < 0.05; and 61% (11/18), p < 0.05] compared to non-diabetic controls [25% (5/20)], suggesting diabetes was a risk factor for AF in mice. We focused on T1D because it is a simpler model system, compared to T2D. We found that enhanced AF risk was prevented in mice with T1D by myocardial CaMKII inhibition due to transgenic expression of a CaMKII inhibitory peptide, AC3-I, suggesting that CaMKII is required for AF in diabetes. Atria from T1D mice showed increased ROS, OGN, and ox-CaMKII. Mice selectively resistant to ox-CaMKII due to knock-in replacement of regulatory domain methionines with valines (MMVV) with T1D were resistant to AF. In contrast, knock-in mice with putative OGN resistant CaMKII (S280A) and wild type (WT) littermate mice with T1D showed similar increases in AF, indicating that OGN-CaMKII did not contribute to AF. WT T1D atrial myocytes, but not MMVV T1D atrial myocytes, showed increased RyR2 Ca2+ leak, delayed afterdepolarizations (DADs) and triggered action potentials, indicating ox-CaMKII coupled ROS to a fundamental proarrhythmic cellular pathway. Addition of an OGN antagonist prevented increased AF, RyR2 Ca2+ leak, DADs and triggered action potentials. Knock-in mice lacking a CaMKII phosphorylation site on RyR2 (S2814A) with T1D were resistant to AF, compared to WT T1D controls, establishing the contribution of CaMKII and RyR2 to AF in this model, and suggesting that OGN is proarrhythmic by a CaMKII independent pathway involving RyR2. Conclusions: These studies establish CaMKII as a critical ROS sensor and proarrhythmic signal in AF, and suggest that ox-CaMKII and OGN converge on RyR2 to promote AF in diabetes. These provide new insights for understanding important clinical phenotypes that currently lack clear mechanisms and adequate therapies.