The invention provides a free-form surface detecting method based on reverse Hartmann detection and relates to the technical field of measurement. Structure position parameters of an optical path system are calibrated; an ideal optical path system model using a detected object as an ideal surface is established; according to the ideal optical path system model, an experiment is performed to obtainwave front aberrations including surface errors and structure errors; parameters of the ideal optical path system model are adjusted to conduct two-step structure error correction based on high-orderand low-order aberration separation and optimization is conducted on the detected data to obtain a standard optical path system model; ray tracing is conducted on the standard optical path system model, wave front aberrations caused by surface errors of the detected surface are detected, and the surface errors of the detected surface are calculated. The technical problem is solved that in the prior art, high-precision free-form surface detection does not have universality. The free-form surface detecting method has the advantages that two-step optimization based on Zernike fitting is conducted on structure errors, structure position errors of the system are effectively eliminated, the detection precision of the system is improved, and detection is universalized.