Predicting phenotypes is a statistical and biotechnical challenge, both in medicine (predicting an illness) and animal breeding (predicting the carcass economical value on a young living animal). High-throughput fine phenotyping is possible using metabolomics, which describes the global metabolic status of an individual, and is the closest to the terminal phenotype. The purpose of this work was to quantify the prediction power of metabolomic profiles for commonly used production phenotypes from a single blood sample from growing pigs. Several statistical approaches were investigated and compared on the basis of cross validation: raw data vs. signal preprocessing (wavelet transformation), with a single-feature selection method. The best results in terms of prediction accuracy were obtained when data were preprocessed using wavelet transformations on the Daubechies basis. The phenotypes related to meat quality were not well predicted because the blood sample was taken some time before slaughter, and slaughter is known to have a strong influence on these traits. By contrast, phenotypes of potential economic interest (e.g., lean meat percentage and ADFI) were well predicted (R(2) = 0.7; P < 0.0001) using metabolomic data.