The high requirements of safe production and quality products have greatly spurred the development of understanding analysis and monitoring technologies of batch processes. However, for the multimode and multiphase batch processes with serious process dynamics, few process monitoring studies have been reported. To address the above issues, a novel process monitoring method based on between-mode similarity evaluation and discriminative information analysis is proposed in this article. First, within each phase, a between-mode similarity method is proposed by classifying samples into two groups, in which the group of mutual samples measures the between-mode similarity and then similar modes within each phase are merged to avoid repetitive modeling. Next, to explain the causes of between-mode differences within each phase, a multimode-Fisher discriminant analysis algorithm is developed to identify discriminative information in a sparse manner, by which discriminative variables contributing most to the discriminative information are obtained. The process understanding is thus enhanced by quantitative similarity evaluation of differences modes and accurate identification of process variables causing between-mode differences. Finally, to effectively improve process monitoring performance, multiple specific monitoring models are constructed for between-mode discriminative information, and a global process monitoring model is, developed for the remaining indiscriminative information, respectively. The proposed method provides detailed insight into the inherent nature of multimode and multiphase batch processes and captures deep process information to enhance one's understanding. The feasibility and performance of the proposed algorithm are illustrated through a typical multimode and multiphase batch process, i.e., injection molding.