Metropolis Monte Carlo (MC) algorithm of the two-dimensional Ising model is used to study the heterogeneous nucleation of protein crystals on rough polymeric surfaces. The theoretical findings are compared to those obtained from classical nucleation theory (CNT), and to experimental data from protein model hen egg white lysozyme (HEWL) crystallized on poly(vinylidene fluoride) or PVDF, poly(dimethylsiloxane) or PDMS and Hyflon homemade membranes. The reduction of the activation energy for the nucleation process on polymeric membranes, predicted to occur at increasing surface roughness, results in a nucleation kinetics that is many orders of magnitude faster than in homogeneous phase. In general, MC stochastic dynamics offers the unique opportunity to investigate the effects of collective molecular aggregation at site level on the nucleation rate and, consequently, allows to identify optimal morphological and structural properties of polymeric membranes for a fine control of the crystallization kinetics.