A new and highly accurate tracking technique, utilizing the basic principle of a Kalman filter, has been described. The technique has the advantage of simplicity and low cost similar to that provided by widely used step-track systems and yet overcomes the major limitations of the steptrack systems, viz, low accuracy and susceptibility to rf carrier-disturbances. This is possible by using a much more accurate algorithm than the step-track algorithm and optimally combining the data with a prediction mode of tracking. The system switches from one mode of tracking to the other, or, optimally combines the two data depending on its confidence on each mode of tracking. For many applications, these features make the system an attractive alternative to the conventional monopulse system. A computer simulation has been used for the study. Assuming typical system parameters, the results predict a 3σ (σ is the standard deviation of tracking error) accuracy of 1/27 of halfpower beamwidth. Under the worst signal conditions, this accuracy is estimated to approach a value of 0.01° or lower.