Metabonomics is an established strategy in the exploration of the effects of various stimuli on the metabolic fingerprint of biofluids. Here, we present an application of (1)H NMR-based metabonomics on the field of exercise biochemistry. Fourteen men were assigned to either of two training programs, which lasted 8 weeks and involved sets of 80-m maximal runs separated by either 10 s or 1 min of rest. Analysis of pre- and postexercise serum samples, both at the beginning and end of training, by (1)H NMR spectroscopy and subsequent multivariate statistical techniques revealed alterations in the levels of 18 metabolites. Validated O-PLS models could classify the samples in regard to exercise, the separation being mainly due to lactate, pyruvate, alanine, leucine, valine, isoleucine, arginine/lysine, glycoprotein acetyls, and an unidentified metabolite resonating at 8.17 ppm. Samples were also classified safely with respect to training, the separation being mainly due to lactate, pyruvate, methylguanidine, citrate, glucose, valine, taurine, trimethylamine N-oxide, choline-containing compounds, histidines, acetoacetate/acetone, glycoprotein acetyls, and lipids. Samples could not be classified according to the duration of the rest interval between sprints. Our findings underline the power of metabonomics to offer new insights into the short- and long-term impact of exercise on metabolism.