Background: Low circulating magnesium (Mg) is associated with an increased risk of developing type 2 diabetes mellitus (T2DM). We aimed to study the performance of a nuclear magnetic resonance (NMR)-based assay that quantifies ionized Mg in EDTA plasma samples and prospectively investigate the association of Mg with the risk of T2DM. Methods: The analytic performance of an NMR-based assay for measuring plasma Mg was evaluated. We studied 5747 subjects free of T2DM at baseline in the Prevention of Renal and Vascular End-stage Disease (PREVEND) study. Results: Passing–Bablok regression analysis, comparing NMR-measured ionized Mg with total Mg measured by the Roche colorimetric assay, produced a correlation of r = 0.90, with a slope of 1.08 (95% CI: 1.00–1.13) and an intercept of 0.02 (95% CI: −0.02–0.08). During a median follow-up period of 11.2 (IQR: 7.7–12.0) years, 289 (5.0%) participants developed T2DM. The association of NMR-measured ionized Mg with T2DM risk was modified by sex (Pinteraction = 0.007). In women, we found an inverse association between Mg and the risk of developing T2DM, independent of adjustment for potential confounders (HR: 1.80; 95% CI: 1.20–2.70). In men, we found no association between Mg and the risk of developing T2DM (HR: 0.90; 95%: 0.67–1.21). Conclusion: Lower NMR-measured plasma ionized Mg was independently associated with a higher risk of developing T2DM in women, but not in men.