The histories of the temperature T of the reaction mass and the agitator current C required for stirring it in a batch polymerization reactor at constant speed are used to develop a soft sensor for online estimation of the state variables, namely, the monomer conversion (x(m)) and the weight average molecular weight (M-w). Methyl methacrylate (MMA) is used as a sample system. Bulk polymerization of MMA is performed in a 1-L stainless steel (SS) reactor instrumented using virtual instrumentation techniques. Experimental data on x(m)(t), M-w(t), C(t), and T(t) are obtained at different values of times t, under various near-isothermal and nonisothermal conditions. A semiempirical model is developed that relates C(t) and T(t) to the state of the polymerization mass. The feasibility of using this as a soft sensor, along with the kinetic model, is demonstrated.