In order to solve the problem of maneuvering target tracking in clutter, a joint probabilistic density description of the target state and target class is given. Using the joint state-class probabilistic density, the predicted measurements density function can be proven as a Gaussian mixture distribution. Based on this attribute, we construct a class-dependent tracking gate for each target class, which achieves more effective association from measurements to tracks. A Gaussian mixture weighted Kalman filter is used in tracking process, where maneuver detection can also be avoided. In simulation, the results with three tracking algorithms are compared, which shows that proposed method here is more effective.