Translational genomics, i.e., the transfer of genetic information from model species to cultivated crops, is on the brink of revolutionizing plant breeding. The recent publication of genomic sequences for several cultivated legumes is also accelerating this process. For pea, recent highthroughput RNA sequencing, and the prospect of a genome sequencing project, will further accelerate the transfer of information from the Medicago truncatula model to the cultivated crop. We have been using genomics approaches with Medicago as a tool to identify key genes determining seed yield and composition in closely related legumes. Analyses of the proteome and transcriptome of the component tissues of the developing seed revealed extensive compartmentalization of gene expression and metabolic activities. Using a TF (Transcription Factor) qRT-PCR platform and the Affymetrix Gene Chip, TFs specific for each seed tissue were identified, along with putative target genes. These TFs have been located on the M. truncatula genetic map and correlations between map positions of TF loci and QTLs for protein quantities and other seed phenotypes were detected. These correlations can be recently confirmed in numerous cases by the existence of similar QTLs at syntenic positions in pea. Two genes, both specifically expressed in the developing endosperm, have received particular attention. One of the genes encodes a DOF class transcription factor, whose mutant phenotype severely affects endosperm development. The second gene encodes an endosperm-specific subtilase (SBT1.1), which affects final seed weight. MiRNAs constitute another level of gene regulation whose importance in the developing seed is beginning to become apparent. We have recently started to compare the sRNA profiles of developing M. truncatula and pea seeds and preliminary results will be presented.