In this paper, prediction for Rice production in Tamilnadu using Bayesian model average and model selection with Zellner's g-prior approach. The methods were performed through MCMC sampling procedures for Bayesian Model Selection and Model Average, Bayesian Predictive Model and Markov Chain Monte Carlo Sampling Method. This posterior distribution can be used to simply select the “best” model BMA performs better than adopting a single best model for prediction. The results in discussed in rice production inclusion Probabilities and p-value normalizing this product, one can infer the PMPs and model weighted posterior distribution for any coefficients.