AbstractThe paper distinguishesbetween optimal and practical collision avoidance systems and selects a set of variables that can be used to compute collision avoidance (C/A) parameters in practical systems. Specifically, the paper studies the effects of sensor errors upon the calculation of the passing distance at CPA, the point‐of‐possible‐collision (PPC), and range over range‐rate (TAU). The possible sources of error are summarized and characterized as deterministic or random, dependent or independent, correlated or decorrelated, and fixed bias. A computer program is described that makes use of a Kalman filter to compute the effects of error upon C/A parameters. The results of computer runs are given that show that most errors have insignificant effects. The others can be reduced or removed if required. A means is provided for estimating a part of the economic return on investment in a computerized collision avoidance