The pseudolinear Kalman filter suffers from more bias and poor tracking accuray in underwater target tracking based on bearings-only information, therefore, a recursive instrumental variable Kalman filter algorithm combined with the bias conpensation algorithm is proposed for bearings-only underwater target tracking, and uncertain process noise covariance is considered. For the bias problem of the pseudolinear Kalman filter, the instrumental variable matrix in the recursive instrumental variable algorithm is adopt to remove the correlation between measurement vectors and pseudolinear noise, as to remove the bias. Simulation results show that the stability and tracking accuracy of the proposed algorithm can be kept in the environment with large underwater noise.