In order to improve the accuracy and the convergence speed of the sphericity error, an improved teaching and learning algorithm is proposed to evaluate the sphericity error. Based on the basic teaching-learning-based optimization, the initial solution quality is improved by logistic chaotic initialization; At the end of each iteration, the interpolation algorithm is applied to the global optimal solution to further improve the search accuracy of the algorithm. Finally, one group of sphericity error algorithm though the measurement data in the related literature is verified the effectiveness of the ITLBO, the test result show that the ITLBO algorithm has advantages in the calculating accuracy and iteration convergence speed, and it is very suitable for the application in the sphericity error evaluation. by genetic algorithm [2], Wen used the immune evolutionary algorithm to the evaluation of sphericity error [3], Hu applied the improved particle swarm optimization method to assess the sphericity error accurately [4], Luo used artificial bee colony algorithm to solve the sphericity error of the measured parts[5], Lei respectively used mesh search algorithm [6] and the geometric approximation algorithm [7] to evaluate the accuracy of sphericity error, Liu constructed the geometrical model of spherical error by the method of chord truncation, and it can obtain the spherical error accurately [8]. In the above sphericity error evaluation algorithm, the intelligent