This volume contains a selection of papers presented at the third Franco-Belgian Meeting of Statisticians held in Rouen (France) on November 25-26, 1982. They reflect the diversity of approaches presently developed in the statistical analysis of Time Series. Some papers present the actual state of research on fundamental problems; others propose new approaches. The first two contributions are devoted to the analysis of time series and to prediction without strong assumptions on the distributions (qualitative harmonic analysis and non-parametric Bayesian prediction). Then various papers are concerned with problems of model selection (choice among linear models, among non-linear models, non-parametric tests, autocorrelation function...). A third group of papers investigates the statistical properties of completely specified parametric models (asymptotic Bayesian analysis, properties of ARMA models, dynamic models with limited dependent variables). The last paper is devoted to a problem in economic analysis relies on the theory of stochastic processes.