This paper investigates the prediction problem in presence of uncertain prior information (UPI) about the parameter of the linear regression model. Accordingly, we propose four predictors, namely the Unrestricted Predictor (URP), the Restricted Predictor (RP), the Preliminary test Predictor (PTP) and finally the Shrinkage Predictor (SP). The bias and the mean square errors (MSE) matrices of the proposed predictors are derived. By studying the MSE matrix criterion, the optimality regions of the predictors are determined under the null as well as the alternative hypothesis.