Background: Quantifying mildly elevated ketone bodies is clinically and pathophysiologically relevant, especially in the context of disease states as well as for monitoring of various diets and exercise regimens. As an alternative assay for measuring ketone bodies in the clinical laboratory, a nuclear magnetic resonance (NMR) spectroscopy-based test was developed for quantification of β-hydroxybutyrate (β-HB), acetoacetate (AcAc) and acetone. Methods: The ketone body assay was evaluated for precision, linearity and stability and method comparisons were performed. In addition, plasma ketone bodies were measured in the Insulin Resistance Atherosclerosis Study (IRAS, n = 1198; 373 type 2 diabetes mellitus (T2DM) subjects). Results: β-HB and AcAc quantified using NMR and mass spectrometry and acetone quantified using NMR and gas chromatography/mass spectrometry were highly correlated (R2 = 0.996, 0.994, and 0.994 for β-HB, AcAc, acetone, respectively). Coefficients of variation (%CVs) for intra- and inter-assay precision ranged from 1.3% to 9.3%, 3.1% to 7.7%, and 3.8% to 9.1%, for β-HB, AcAc and acetone, respectively. In the IRAS, ketone bodies were elevated in subjects with T2DM versus non-diabetic individuals (p = 0.011 to ≤0.001). Age- and sex-adjusted multivariable linear regression analysis revealed that total ketone bodies and β-HB were associated directly with free fatty acids (FFAs) and T2DM and inversely with triglycerides and insulin resistance as measured by the Lipoprotein Insulin Resistance Index. Conclusions: Concentrations of the three main ketone bodies can be determined by NMR with good clinical performance, are elevated in T2DM and are inversely associated with triglycerides and insulin resistance.