Abstract: A method to estimate spacecraft attitude using Risk Sensitive Filter was presented based on Cubature Rules(CF-RSF). The objective was to solve the uncertain problem of system model and noise in complex condition. Introducing risk sensitive function, this scheme overcame the poor robustness and even divergence of filter caused by system uncertainty, and solved nonlinear integral problem using cubature rules. The filter accuracy and robustness on system uncertainty were both improved, and the computational complexity was decreased. The simulation shows that the developed algorithm is effective.