Time series often have patterns that form a basis for comparing them or classifying them into groups. Pattern recognition of time series arises in a number of practical situations. Procedures for the comparison and classification of univariate stationary series already exist in the literature. A famous application is the comparison and classification of earthquake and nuclear explosion waveforms - Shumway (1982). In this paper we present procedures to compare and classifi,stationary multivariate time series. Simulations studies show that the procedures perform fairly well for reasonably long series.