Flavor quality is one of the most difficult traits to select in plant breeding programs due to the large number of sensory panelists required, the small number of samples that can be evaluated per day, and the subjectivity of the results. Using sweetpotato [Ipomoea batatas (L.) Lam.] as a model, clones exhibiting distinctly different flavors were analyzed for sugars, nonvolatile acids, and aroma chemistry to identify the critical flavor components. Differences in sugars, sucrose equivalents, nonvolatile acids, and 19 odor-active compounds were identified that accounted for differences in flavor among the clones. Using the intensity of the aroma per microliter for each of the 17 most important aroma-active compounds (maltol, 5-methyl-2-furfural, 2-acetyl furan, 3-furaldehyde, 2-furmethanol, benzaldehyde, phenylacetaldehyde, β-ionone, 1,2,4-trimethyl benzene, 2-pentyl furan, 2,4-decadienal, 2,4-nonadienal, linalool, geraniol, cyperene, α-copane and a sesquiterpene) and the relative sweetness of individual sugars × their respective concentrations, multivariate (principal component and cluster) analysis allowed accurate classification of the clones according to flavor type without sensory analysis. The level of precision was such that sweetness, starch hydrolysis potential, and the concentration of β-carotene could be accurately predicted by quantifying specific volatiles. Analytical assessment of flavor would greatly facilitate the accurate evaluation of large numbers of progeny, the simultaneous selection of multiple flavor types, and the development of superior new cultivars for a wide cross-section of food crops.