In this article, the problem of obtaining Bayesian prediction bounds for some certain order statistics in samples from the (Gomp(α, β)) distribution has been studied in the presence of outlier arising from different members of the same family of distributions. By using a bivariate prior density for α and β a single outlier of types β β0 and β + β0 are obtained. Markov Chain Monte carlo (MCMC) has been used to obtain Bayesian prediction intervals for both single outliers of types β β0 and β + β0. Numerical examples are used to illustrate the procedure.