Significant economic benefits can be derived from reducing the variability of product properties around their specification targets in batch reactors. End-use properties in batch reactors using dynamic models that relate the operating conditions to the end-use properties are related through a static nonlinear model to the molecular properties of the product and are controlled directly. Their values are required to lie in a desired target region in the property space. The control of end-use product properties is formulated as a nonlinear model predictive control problem, and an efficient numerical technique using successive linearizations is utilized for the solution. A parameter adaptive extended Kalman filter is used for state estimation of the molecular properties that are not measured directly. The emulsion polymerization of styrene is chosen as a case study, and an approximate mathematical model is developed. The model is detailed enough to predict the end-use properties, which depend on both the molecular weight and particle-size distributions of the product. The end-use product properties, such as tensile strength and melt index, are controlled in desired target regions by adding a monomer and a chain transfer agent, as well as the flow rate of the coolant, as the manipulated variables.