A new algorithm of evaluating conicity error, named as geometrical optimization searching algorithm (GOSA), has been presented, according to the geometric characteristics of conicity error. The principle and steps of using this algorithm to solve the conicity error was described in detail. The mathematical formula and program flowchart were given. The optimization method and linearization method and uniform sampling did not adopt for this algorithm. The principle of the algorithm is that varying the position of the supposed ideal conical surface repeatedly according to pre-established transfer mode, calculating the distances between all measuring points and the supposed ideal conical surface and comparing the extreme differences among the distances at the same time, and thus the minimum zone method of conicity error is obtained. The experimental results show that the conicity error can be evaluated using this algorithm effectively and exactly.