We investigate the effect of the presence of reactive impurities during copolymerization on mathematical model discrimination and determine if impurities adversely affect the model discrimination process, depending on the criteria and response(s) used. Nonlinear mechanistic process models are employed in a multiresponse statistical discrimination scenario. The investigation also considers, via several case studies, the extremely interesting question of whether impurity effects could lead to incorrect choices about the copolymerization mechanism itself (terminal versus penultimate model). The answers to the above questions are that reactive impurities do affect the model discrimination process and, what is more, they may mislead researchers as to the choice of the copolymerization mechanism. The latter unexpected result is further illustrated with experimental data from the literature.