A model composed of a synthesis of the nonlinear Cahn-Hilliard and Flory-Huggins theories for spinodal decomposition (SD) and a second-order rate equation for the self-condensation of a trifunctional monomer is presented and used to analyze polymerization-induced phase separation (PIPS). The numerical results replicate frequently reported experimental observations on the PIPS of a binary monomer-solvent solution. These observations include a transient periodic concentration spatial profile with a wavelength that decreases with increasing rate constant. In addition, the time evolution of the maximum value of the structure factor exhibits an exponential growth during the early stage, but then slows down in the intermediate stage of SD. Computational analysis shows that, in the PIPS method, the wavelength of the phase-separated structure depends on the complex interaction between the competing polymerization and phase separation processes. The effects of these two competing processes on the characteristic time and length scales of the phase separation phenomena depend on the magnitudes of a scaled diffusion coefficient D for phase separation and a scaled rate constant K-1 for polymerization. As D increases, the dominant dimensionless wavenumber k(m)* also increases, but the phase separation lag time decreases. Similarly, as K-1 increases, k(m)* also increases, but the polymerization lag time decreases. On the basis of these two dimensionless parameters, the dominant wavelength selection mechanism in the PIPS process is identified, which enables the control of morphology during the PIPS phenomena.