This paper presents an efficient suboptimal decision algorithm for associating (correlating) sensor data with stored tracks in a real-time track-while-scan surveillance system. For the first time, the performance and computational burden of such an algorithm is derived analytically. The results provide considerable insight into the correlation process, and permit the tracking system designer to select and trade off various tracking and correlation parameters, to achieve acceptable performance and implementation cost without resorting to detailed and costly computer simulations required by previous ad hoc approaches to the problem.