Abstract: 【Objective】This study aimed to investigate effects of different magnesium concentrations on physiological characteristics of leaves at different leaf positions of the flue-cured tobacco variety Yunyan 87， conduct inversion analysis of relevant indexes based on hyperspectral technology， evaluate the feasibility of using this technology for rapid detection of physiological indexes of tobacco leaves， thereby providing a theoretical basis for precise regulation of magnesium nu-trients in flue-cured tobacco and improvement of tobacco leaf quality.【Method】The flue-cured tobacco variety Yunyan 87 was used as experimental materials in a hydroponic experiment. Three magnesium concentration treatments were set up： conventional magnesium supply （CK， 5 mmol/L）， magnesium deficiency treatment （T1， 0 mmol/L）， and high magnesium treatment （T2， 25 mmol/L）. On 30， 40， and 50 d of hydroponic culture， representative tobacco plants were selected to measure the photosynthetic indexes， chlorophyll content， antioxidant enzyme activity， and malondialdehyde （MDA） content of leaves at the 3^rd， 8^th， and 13^th positions from the bottom up. Hyperspectral data of leaves were also collected， and the partial least squares regression （PLSR） model was used to invert the relevant indicators.【Result】Compared with CK， the response characteristics of Yunyan 87 to magnesium deficiency and high magnesium were manifested as a decrease in chlorophyll content in leaves at the 3^rd， 8^th， and 13^th positions， an increase in the activities of superoxide dismutase （SOD） and peroxidase （POD） as well as MDA content， a decrease in catalase （CAT） activity， and inhibited photosynthesis， with the most pronounced effects observed in the lower leaves under magnesium deficiency treatment. Hyperspectral curves of leaves treated with different magnesium concentrations showed overall consistent trend， but spectrum showed different reflectance. Based on differences in spectral curves， full spectrum PLSR models were established based on chlorophyll content， antioxidant enzyme activity， MDA content， and photosynthetic indexes. The model prediction results showed that the optimal pretreatment method for chlorophyll content， POD activity， and CAT activity was standard normal variate （SNV）； the optimal pretreatment methods for SOD activity， net photosynthetic rate （P_n）， intercellular CO₂ concentration （C_i）， and transpiration rate （T_r） was Savitzky-Golay filter smoothing （S-G）； the optimal pretreatment methods for MDA content and stomatal conductance （G_s） was multivariate scattering correction （MSC）. Among them， the prediction accuracy for chlorophyll content was the highest， with a prediction set correlation coefficient （r_p） of 0.7833.【Conclusion】Magnesium deficiency or excess affects the growth and development of flue-cured tobacco， with magnesium deficiency showing a more significant impact， and leaves at lower leaf positions are more sensitive to magnesium deficiency. Among the inversion models established based on hyperspectral technology， the chlorophyll content inversion model exhibits the highest prediction accuracy， which is expected to realize rapid monitoring and evaluation of magnesium nutrients in flue-cured tobacco.