In this paper we use the Gibbs sampling algorithm to present a Bayesian analysis to multiplicative seasonal autoregressive moving average (SARMA) models, considering both estimation and prediction problems.