The pathologic condition of a patient presenting a metabolic disease can change rapidly, and a variety of pathologic conditions are possible. Plasma Fourier-transform infrared (FT-IR) spectra were used to differentiate patients with type 1 diabetes, healthy subjects, and endurance-trained rowers. Analytic and classification methods that use the same plasma FT-IR spectra are described. Complete spectra (4000 to 500 cm(-1)) classifications led to a differentiation between most patients with type 1 diabetes and other subjects but not between control and trained subjects. Classification of defined absorption regions of spectra allowed different metabolic distinctions between populations. These were performed on the amide I and II absorption regions of proteins (1720 to 1480 cm(-1)); on the nu=CH, nu(as)CH(2), and nu(as)CH(3) absorption regions of lipids (3020 to 2880 cm(-1)); and on the nuC-O absorption region of saccharides (1300 to 900 cm(-1)). A classification that uses a combination of four absorption regions-nu=CH (3020 to 3000 cm(-1)), nu(as)CH(3) (3000 to 2950 cm(-1)), nuC-O (amide I: 1720 to 1600 cm(-1)), and nuC-O (carbonyle: 1300 to 900 cm(-1))-led to the formation of three exclusive clusters that comprised the defined populations. FT-IR spectroscopy is an exciting technique that allows a versatile approach to biologic samples from which analytic and statistical methods might be used for metabolic profile characterization and evaluation.