Abstract: Atractylodes macrocephala Koidz. is a perennial herb belonging to Compositae family, which is fond of cool climate regions. Rhizome dried for over 2 years are used for a series of medicinal functions. It is a top medicinal herb in Zhejiang Province, China. At the early stage of rhizome enlargement, A. macrocephala has a certain tolerance to mild soil drought, but excessive drought can inhibit rhizome enlargement and accumulation of constituent chemicals. Increasing degree of drought stress could slow down growth, inhibit rhizome enlargement and limit yield. In order to provide reference for drought stress control and cultivation of drought-resistant varieties, biennial A. macrocephala were planted under different drought stress. Spectral reflectance of A. macrocephala leaves determined by UniSpec-SC spectrum analyzer and combined with photosynthetic pigment contents were used to explain the response of spectral characteristics under drought stress. The results showed that spectral reflectance increased in the visible region (400-750 nm) with increasing drought stress. This indicated that the absorption and utilization ability of light energy decreased under increased drought stress. However, spectral reflectance gradually stabilized in the near-infrared band 750-1 000 nm. The reflectance of all the leaves under drought stress was lower than that of the control at 1 000 nm. The difference in spectrum of 680-750 nm was significant, which was correlated with chlorophyll content in 700-750 nm. This band could be used to monitor whether A. macrocephala was affected by drought. The contents of photosynthetic pigments increased initially and then decreased with the increased drought stress. It indicated that a suitable degree of drought was good for growth. Most of the spectral parameters were significantly correlated with pigments contents (P < 0.05). Spectral parameters of mCRI, PSNDb, red-edge position (λ_red), red-edge amplitude (D_λred) and red-edge area (S_red) were significantly correlated with leaf pigments contents (P < 0.01), which could be used to diagnose drought indicators. In summary, differential spectrum of 680-750 nm could be used to detect drought impact on A. macrocephala. Red-edge parameters, carotenoid reflectance indexes and pigment specific normalized difference may be used as indicators to diagnose drought stress degree of A. macrocephala. This conclusion not only provided a reference for the study of high-spectrum plant research, but also provided theoretical basis and technical support for the application of spectrum diagnosis of A. macrocephala in drought stress analysis.