A novel non-null annular subaperture stitching interferometry (NASSI) is proposed for steep aspheric test. It combines the non-null test and annular subaperture stitching method, in which a partial null lens is employed as an alternative to the transmission sphere, to generate aspherical wavefront as the reference. The aspherical wavefront has a better performance in matching the local slope of subapertures. This approach greatly reduces the subaperture number required to cover the full-aperture of test surface. On one hand, subapertures are broadened and thus the stitching accuracy increased. For another, accumulated error is decreased. Based on the system modeling, the retrace error can be corrected for each subaperture individually and the testing accuracy is further improved. A numerical simulation exhibits the high theoretical accuracy of the NASSI method, in which a high-order aspheric of 25 μm asphericity is tested. Comparing experiments with Zygo VerifireTMAsphere interferometer for paraboloid, whose diameter is 101 mm, are performed. The peak-valley error is better than λ/20, the root-mean-square error is better than λ/100. The high accuracy and good repeatability of NASSI are presented.