In this paper we use the Gibbs sampling algorithm to present a Bayesian prediction of multiplicative triple seasonal autoregressive (TSAR) models.