For complex industrial processes with multiple operating conditions and dynamic characteristics, the traditional dynamic process monitoring techniques (e.g., canonical variate analysis, CVA) are not well-suited, because the fundamental assumption that the operating data follow a unimodal Gaussian distribution usually becomes invalid. In this article, a novel mixture canonical variate analysis (MCVA) model is proposed to model and monitor multimode dynamic processes. First, the augmented process data are assumed to be many different clusters, each of which corresponds to an operating mode and can be characterized by a Gaussian component. Then, singular value decomposition of the covariance matrices is implemented in each Gaussian cluster and the corresponding canonical variates are obtained. For process monitoring purposes, the local statistics in each cluster are calculated and the integrated monitoring indices are obtained in a probabilistic manner. The validity and effectiveness of the proposed monitoring approach are illustrated through two examples: a numerical process and the Tennessee Eastman challenge problem. The comparison of monitoring results demonstrates that the proposed approach is superior to conventional CVA and Gaussian mixture models (GMM).