A starved emulsion polymerization of styrene was investigated, with the aim of controlling the molecular weights of the produced polymer. The reactor was operated under a starved feed of a mixture of monomer and chain transfer agent (or “modifier”). Tert-dodecyl and tert-nonyl mercaptans were the employed modifiers. The starved operation produced a constant “most probable” molecular weight distribution along the polymerization. A mathematical model was developed to help interpret the effect of the modifier chain length. To adjust some of the model parameters, two batch experiments were carried out. In the starved polymerization with tert-docecyl mercaptan, a mass transfer resistance to the modifier was required to fit the observed molecular weights. This extra mass transfer resistance could be neglected in the case of the (more water soluble) tert-nonyl mercaptan, however. The developed model also successfully predicts the batch experiments of Harelle et al. (J. Appl. Polym. Sci. 1994, 52, 1105−1113).