A new method of increasing manoeuvring target tracking performance is presented. The target manoeuvring is modelled as process white gaussian noise so that the system equations to which optimal filtering theory applies can be obtained. A multi-model adaptive Kalman filter is developed to estimate the target positions, based on process noise covariance matching and multi-model measurement noise. In the conventional adaptive filter, only one of process noise and measurement noise covariance matching can be done in a program. In this algorithm, the tracking filter estimates the covariance of process noise and measurement noise simultaneously to increase the accuracy of the conventional real-time tracking filter. Finally, an illustrative numerical example is included.