During in vitro differentiation, pluripotent stem cells undergo extensive remodeling of their gene expression. While studied extensively at the transcriptome level, much less is known about protein dynamics, which can differ significantly from their mRNA counterparts. Here, we present genome-wide dynamic measurements of mRNA and protein levels during differentiation of embryonic stem cells (ESCs). We reveal pervasive discordance, which can be largely understood as a dynamic imbalance due to delayed protein synthesis and degradation. Through a combination of systematic classification and kinetic modeling, we connect modes of regulation at the protein level to the function of specific gene sets in differentiation. We further show that our kinetic model can be applied to single-cell transcriptomics data to predict protein levels in differentiated cell types. In conclusion, our comprehensive data set, easily accessible through a web application, is a valuable resource for the discovery of protein-level regulation in ESC differentiation.