The problem of simultaneous control of composition and molecular weight of copolymers produced in semicontinuous emulsion polymerization is addressed. A first-principles model of emulsion copolymerization is used to produce optimal open-loop policies, which are then experimentally tested on the system methyl methacrylate/vinyl acetate. In contrast to earlier work in the control of semicontinuous emulsion copolymerization, the control problem addressed has multiple inputs and multiple outputs, and the experimental application is free of empirical correlations and performed on an unseeded system. Optimal open-loop policies are calculated efficiently by optimizing properties of the instantaneous polymer being formed. The experimental results confirm the feasibility of these techniques. A model-predictive control scheme, in which optimal policies are re-computed on-line to account for disturbances, is proposed. The scheme, which exploits the advantages of the efficient calculation of optimal policies, is tested by simulation showing good performance.