We have previously developed dynamic flux balance models for prediction of cellular growth and metabolic product formation rates in Saccharomyces cerevisiae batch and fed-batch cell cultures. These models couple steady-state stoichiometric balances on intracellular metabolites with dynamic extracellular balances on biomass, substrates, and metabolic byproducts through time-varying substrate uptake rates. In this contribution, we present the results of Saccharomyces cerevisiae fermentation experiments designed to evaluate the dynamic flux balance model predictions. We utilized a compartmentalized, genome-scale metabolic network to describe intracellular metabolism and simple Michaelis-Menten kinetics for the glucose and oxygen uptake rates. Kinetic parameters for the substrate uptake and the intracellular stoichiometric coefficients representing growth and non-growth associated energy requirements were estimated by nonlinear least squares optimization. A series of batch and fed-batch experiments demonstrated that the dynamic flux balance model was able to produce accurate substrate, biomass, and extracellular product concentration profile predictions over a wide range of fermentation conditions. The model was incorporated with a bi-level dynamic optimization scheme to compute fed-batch operating policies for optimal ethanol production in batch and fed-batch cultures. Experimental implementation of the optimal policies showed good agreement with the in silico predictions.