To improve the filtering precision under complex noise condition,a new Square-root Cubature Cost-Reference Particle Filter (SCCRPF) is proposed based on the Square-root Cubature Kalman Filter (SCKF) and the Cost-Reference Particle Filter (CRPF).The proposed filter updates the prior distribution function with the latest measured information and SCKF,and thereby generates the importance density function for CRPF.The new filter not only reserves the precision advantage of SCKF in filtering nonlinear systems,but also possesses the filtering precision of CRPF for dealing with systems with an unknown noise assumption.Simulation results show that:the filtering precision of SCCRPF is higher than that of Square-root Cubature Particle Filter (SCPF) for a system with unknown noise assumption,and is higher than that of CRPF for a system with known noise assumption.