The detection of skin's structure lays the foundation for personalized laser surgery of vascular skin disease, which can be noninvasively achieved by diffuse reflectance spectroscopy (DRS). A two‐step inverse Monte Carlo radiation method based on DRS under two source‐detector separations was proposed to quantify the skin structure, including chromophore concentration (melanin fm and hemoglobin fb), epidermal thickness tepi, average vessel diameter Dves, depth dpws and thickness tpws of the vascular layer for diseased skin. The method fitted the simulated DRS to the measured DRS iteratively, differences between which were described by a specific objective function to amplify blood absorption at 500–600 nm, and Dves, dpws, and tpws were estimated based on fm, fb, and tpws fitted in the first step. The results showed that the two‐step method dramatically improve the inversion accuracy with mean errors of fm, fb, tpws, and dpws less than 5%.