A new class of dynamic models for stationary time series is presented. It is a natural dynamic generalization of the well-known Factor Analysis Model widely used in Statistics. Factor Analysis models of time series are also related to dynalaic Errors-in-Variables models discussed in the recent literature. They provide simple mathematical schemes for the identification of multivariate time series which a-void the unjustified introduction of causality relations among the variables, as for example subsumed by conventional ARNAX models.