經驗分布在機率論及統計學都很重要，經驗分布為非母數統計之重要工具，本文研究經驗分布在母數統計之表現，以常態、柯西和Levy等穩定型分布為例，用經驗分布作參數估計並和常用統計量比較。 Empirical distribution has important theoretical property like Glivenko-Cantelli theorem from which statistical inference for distribution function is possible. Empirical distribution is optimal in nonparametric framework. However, it is unknown if empirical distribution is optimal in parametric models. In this paper, we will investigate the eﬃciency of empirical distribution in parametric framework. The distributions considered include normal, Cauchy and Levy that are all stable.