In this paper an extension to previous research into dynamic process performance monitoring using the generic dissimilarity measure (GDM) is described. The proposed method, wavelet-based GDM, combines the ability of wavelets to extract deterministic features and approximately decorrelate autocorrelated process data with that of the GDM. The GDM realizes the monitoring of changes in the time series distribution of the process data with the degree of dissimilarity defining the monitoring metric. The objective of the wavelet aspect of the analysis is to approximate the true process characteristics and to address the correlation structure present in the data whilst the GDM is the basis of the process performance monitoring step that is implemented using a moving window approach. A comprehensive analysis of data simulated from the Tennessee Eastman process indicates that the proposed method is effective in detecting process changes and faults of various magnitudes with little or no time delay. Furthermore a comparative study with three other methods, standard PCA, dynamic PCA and multiscale PCA, shows that the proposed approach exhibits superior performance in terms of false alarm rate and time to fault detection.