Changes of attribute weights directly affect the results of grey incidence analysis for advancing the stability of results on grey relation decision-making. This chapter proposes an optimization method of grey relation analysis based on the minimum sensitivity of attribute weights. The main idea is as follows: First, we certify that sensitivity satisfies the properties of outer measure, explains that it is reasonable to use the sensitivity to measure the stability of grey relation analysis. Then the conditions and properties are given when the changes of attribute weights in grey relation analysis are effective. On the purpose of getting the minimum sensitivity we construct a multi-objective quadratic programming model based on objective weighting method and the partial preference information of the decision maker. The particle swarm optimization (PSO) algorithm is used to solve the model, and then we get grey relation analysis based on the weight. Finally, this method is used to evaluate R&D human resources.