Microgrids with a high penetration of renewable energy sources face pronounced source–load uncertainties. Existing uncertainty-driven scenario reduction methods often fail to incorporate the characteristics of the optimization model and exhibit insufficient risk-aversion capability. This paper proposes an objective-oriented scenario reduction method. First, a two-stage day-ahead and intra-day optimal scheduling model is formulated for the microgrid under uncertainty. Next, the distance between scenarios in the objective space is defined and analytically derived. An objective-oriented scenario clustering method based on a proxy scenario set is then developed, and scenario weights are reassigned according to the operational risk of each cluster to obtain representative scenarios and risk-aversion weights aligned with the optimization objective. Finally, simulations on microgrids derived from the IEEE 33-bus and IEEE 123-bus systems demonstrate that the proposed method effectively enables targeted selection of representative scenarios, reduces microgrid operating costs, and enhances the robustness of dispatch decisions.