An aberration estimating method using a steepest descent method is configured to estimate, as an aberration of a test optical system, an aberration when a predetermined evaluation function becomes less than or equal to a permissible value. The aberration estimating method comprising the step of updating the aberration with a sum of a current aberration and a first derivative of the evaluation function by the aberration when the evaluation function is larger than the permissible value. The aberration is an aberration of an entire pupil plane of the test optical system. The updating step includes calculating the first derivative by Fourier-transforming the difference instead of an integration at coordinates of respective points on an image plane of the test optical system.