Distributed parameter estimation is more practical in wireless sensor networks, as it has less communication overhead and is robust in large scale sensor networks. To solve the state estimation problem of nonlinear and non-Gaussian system, we propose a distributed cubature Kalman particle filter, which use cubature Kalman filter to generate the importance proposal distribution of particle filter, it can solve the particle degradation problem and improve the estimation accuracy. The system merge the estimations via weighted linear combination, the weighting factor is obtained by linear minimum variance criterion. Simulation shows that the algorithm improves the accuracy of estimation than the single sensor subsystem.