Volatile organic compounds (VOC) analysis, thanks to marked recent technical advances, is emerging as one of the most promising tools for the development of innovative research and clinical methodologies. The general premise to this proposal is that in states of altered metabolism, such as type 1 diabetes mellitus (T1DM) or type 2 diabetes mellitus (T2DM), important changes may occur in patterns of exhaled VOCs, reflecting endogenous changes at the tissue and cellular levels. In this dissertation project, we performed complex in vivo metabolic clamping experiments on both healthy and diabetic subjects, during which we collected matched breath, room air, and blood samples. Concentrations of ~100 volatile organic compounds (VOCs) were then identified and quantified by gas chromatography and mass spectrometry. Corresponding plasma glucose was measured by a high-precision glucometer; insulin and lipid concentrations were measured by ELISA. We then developed predictive algorithms allowing the non-invasive, breath-based measurement of key metabolic variables (plasma glucose and lipids) using multilinear regression analysis. Furthermore, we developed several metrics to assess the stability of VOCs and predictive algorithms across various metabolic perturbations over time. The general rationale for these studies is that current methodologies used to measure critical metabolic variables (plasma glucose, insulin, cholesterol panel) require invasive blood draws, are expensive, and can be painful. The broader, downstream goal of the project is to translate this technology into practical clinical devices that will simplify monitoring of these blood variables. Such alternative, non-invasive testing strategies will likely greatly improve prevention, diagnosis and everyday management of diabetes mellitus and other related conditions (e.g. obesity, cardiovascular disease).