In this paper, we suggest a new efficient method, that yields two new point predictors, for predicting future order statistics, when the sample size remains fixed and when it is enlarged. The two suggested point predictors are constructed based on some characterization properties of order statistics via their distributions. A comprehensive simulation study is conducted on several distributions, among them is a mixture distribution, to evaluate the performance of the suggested predictors. Two examples of lifetime real data are analyzed in which several methods of the point predictions are compared.