In this article, our aim is to outline basic properties of best linear unbiased prediction (BLUP). We first introduce the general linear model , where is the covariance matrix and the expectation of the response variable . We then let be an unobservable random vector generated by , where the covariance matrix between the random error terms and is known. Our main goal is to predict linearly the unobservable on the basis of the observable . BLUP involves finding a matrix so that is unbiased for , and the covariance matrix of the prediction error is minimal in the Löwner sense. Then, via setting in the linear mixed model , we show how prediction of can be used to provide BLUPs of the random effects, and hence of . Empirical best linear unbiased prediction (EBLUP), used when covariances are estimated rather than known, is then outlined.