With the definition of the geometry error and the geometric characteristics of sphericity error, a sphericity error evaluation method with the name of mesh search algorithm was presented. The principle of the algorithm was that with the least-square as the initial reference point, and a series of mesh points were collocated around the periphery of the least-square mad self-alignment according to the certain rules, the radius values of all the measured points were calculated by using each mesh point as the ideal center. By comparison and judgment to obtain corresponding evaluation methods (minimum circumscribed sphere (MCS), maximum inscribed sphere (MIS), minimum zone sphere (MZS)).The principle and the realization process of the mesh search algorithm were described in detail. The examples results show that the sphericity error can be evaluated effectively and exactly by using this algorithm.