The manufacture of emulsion polymers with broad and bimodal particle size distribution (PSD) through in situ particle nucleation in semibatch reactors is difficult because of the sensitivity of particle nucleation phenomena to variations in reactor conditions, impurities, and surfactant and initiator properties. In this paper, we present several control strategies based on the use of readily available online and offline measurements to predict the final PSD and, if necessary, to compute mid-course corrections. Partial least squares (PLS) models are used to extract the necessary information from different sets of measurements, to predict the final PSD, and to define a control region, outside of which mid-course control actions are deemed necessary. Using a simulated styrene emulsion polymerization process as an example, these control strategies are shown to be highly effective and very practical for industrial implementation.