This paper addresses the issue of modelling and control of multivariable chemical process systems using the dynamic version of a popular multivariate statistical technique, namely, Projection to Latent Structures (Partial Least Squares or PLS). Discrete input-output data is utilized to construct a projection based dynamic model that captures the dominant features of the process under study. The structure of the resulting model enables the synthesis of a multi-loop control system. In addition, we also present the design of feedforward control for multivariable systems using the dynamic PLS framework. Three case studies will be used to illustrate the modelling and control of multivariable linear and nonlinear systems using the suggested approach.