Protein and amino acid contents of whole lentils seeds were predicted using Near Infrared Spectroscopy. Predictions were made using a partial least squares (PLS) model from the R package, ‘pls’.  The optimal number of components used to predict protein and amino acid contents were selected using a permutation approach.  The permutation approach finds the number of components with the minimum root mean squared error of prediction, then removes components stepwise to find the minimum number of components without impacting model performance.  This experiment examined LR-68, which is an interspecific cross between cultivated lentil (Lens culinaris) and a wild relative (Lens orientalis). This population contains a diverse set of seed characteristics, including protein and amino acids. A previously developed genetic linkage map will be used to detect regions of the genome responsible for protein and amino acid content of seeds.