In the present paper, the current status of covariance applications to fast reactor analysis and design is summarized with actual examples. The covariance applications are classified into three fields. First, covariances are used to quantify the uncertainty of nuclear core parameters such as criticality, control rod worth, reaction rate ratio, power distribution, sodium void reactivity, etc. The special features of the Japanese case are to include both the burnup-related parameters such as burnup reactivity loss or fuel composition changes, and the temperature-related parameter, that is, the Doppler reactivity. Second, covariances are used to select the important nuclides, reactions and energy ranges which are dominant to the uncertainty of core parameters, and to give nuclear scientists recommendations to improve the accuracy of the data. Finally, covariances are used to improve the accuracy of core design values by adopting the integral data such as the critical experiments and the power reactor operation data. The way toward improvement is classified into the conventional E/C (Experiment/Calculation) bias method and the more comprehensive cross section adjustment based on the Bayesian theorem and the generalized least square method. In Japan, an adjusted group-constant set, ADJ2000R, is now being used in the design work for future fast reactors.