Wireless sensor networks are highly suitable for target tracking. In such networks, node scheduling is critical in terms of energy consumption and tracking accuracy. A wireless sensor network consisting of nodes with multiple bearings-only sensing units is considered in this study and an adaptive approach for the scheduling of such a network is presented. The proposed method continually determines the minimum number of measurements each node should perform in order to keep the tracking error below a certain limit. An extended Kalman filter is utilized for target tracking. Tracking accuracy is formulated based on the nonlinear A-optimality measure. The node scheduling problem is reformulated as a linear binary mixed integer programming problem and solved in an online manner. Simulation results show that the proposed method results in less RMS position error than the nearest node selection method, especially at the beginning of tracking when there is more uncertainty about the target state. With the same amount of energy consumption, the number of divergent tracks in this method is much fewer in comparison with those of the nearest node selection method.