Cooperative positioning refers to a new positioning technique that receivers share critical positioning information with each other by communication system so as to achieve relative localization. The receivers estimated their position with available secondary nodes which included the localized neighbor receivers and the visible satellites. The cooperation of receivers resulted in the increasing of the available secondary nodes for position estimation, so the calculated amount and the computing time were increased. With the increasing of available secondary nodes, the receivers may fail to estimate their position in real-time. A new node selection algorithm called improved quasi-optimal node selection algorithm was proposed. The nodes to be selected obtained an importance value by calculating the unit vector differences between nodes, and the ones with the maximum importance value were selected as the secondary nodes in each iterate. The new algorithm was compared with four typical node selection algorithms. Simulation result shows that the nodes selected by the proposed algorithm are very close to the ones selected by the optimal algorithm, but the former algorithm costs less time. And the comprehensive performance of the new algorithm is better than the typical node selection algorithm.