Abstract: Accurate assessment of crop nitrogen nutrition status is essential for precision fertilization and sustainable agricultural development. This study aimed to develop rapid and non-destructive diagnostic models to predict aboveground plant nitrogen concentration (PNC), leaf area index (LAI) and nitrogen nutrition index (NNI) of sorghum by means of unmanned aerial vehicle (UAV)-based hyperspectral remote sensing. Further the built models were validated across different cultivars and regions. Field experiments were conducted in 2024 at two sites (Shanyin and Yuci) in Shanxi Province. At Shanyin, the field experiment comprised six nitrogen application rates (0, 30, 75, 150, 225, and 300 kg·hm⁻²) and three sorghum varieties (Jinzao 5564, Jinzao 5577, and Jinza 22). At the Yuci site, long-term fertilization experiments, using Jinza 22, included five treatments: no fertilization (CK), nitrogen omission (PK), phosphorus omission (NK), potassium omission (NP), and complete NPK fertilization (NPK), which were initiated in 2011. During sorghum vegetative growth period, plants samples were collected at 41, 50, 63, and 72 days after emergence at Shanyin, and at 28, 37, and 59 days after emergence at Yuci. Simultaneously, a drone equipped with a hyperspectral sensor was utilized to acquire remote sensing data over the study area. Aboveground PNC and LAI were determined immediately after sampling. NNI was calculated based on PNC and critical N concentration (Nc), which was estimated using a critical nitrogen dilution curve derived from LAI established from previous research. Competitive adaptive reweighted sampling (CARS), uninformative variable elimination (UVE) and successive projections algorithm (SPA) were applied to raw spectral reflectance (R) and first-derivative reflectance spectra (FDR) to identify characteristic spectral bands. Partial least squares regression (PLSR) models were developed based on data from Jinzao 5564 and Jinzao 5577. Jinza 22 was used for cultivar validation, and Yuci experiment served for regional validation. Model performance was evaluated using the coefficient of determination (R²) and root mean square error (RMSE). Results showed that nitrogen application significantly influenced aboveground PNC, LAI, and NNI (P < 0.05). PNC gradually declined during crop development, whereas LAI increased over time. N application significantly promoted NNI, which researched 1 at an application rate of 150 kg/hm². In the long-term experiment conducted at Yuci site, NNI values researched 1 when nitrogen was applied (e. g. NK, NP and NPK treatments); whereas it ranged from 0.46 to 0.67 for both the CK and PK treatments. Sensitive spectral bands were selected from R and FDR to predict PNC, LAI, and NNI using CARS, UVE and SPA, respectively. Regardless of whether R or FDR were used, the feature bands sensitive to PNC, LAI and NNI were mainly concentrated in the range from red to near-infrared. However, in terms of the same predicted indicator, the sensitive feature bands were variable brought from feature selection algorithm. The PLSR model based on raw spectral reflectance combined with CARS-selected bands achieved the best performance for predicting PNC (R² = 0.907, RMSE = 0.163%) and LAI (R² = 0.940, RMSE = 0.480) on the test set. In cross-cultivar validation, the R² of R-CARS model for PNC and LAI were 0.883 and 0.878, respectively. For cross-regional validation, Meanwhile the corresponding R² values were 0.643 and 0.911. The PLSR model using raw spectral reflectance combined with UVE-selected bands provided the highest accuracy for NNI prediction, with an R² of 0.754 and RMSE of 0.086. The R² values for cultivar and regional validation were 0.666 and 0.529, respectively. Although cross-regional prediction accuracy for PNC and NNI was lower than that of the test set, the spatiotemporal distribution patterns of the predicted PNC, LAI, and NNI remained consistent with nitrogen application gradients and the dynamics of nitrogen nutrition across growth stages. In conclusion, the developed models effectively captured variations in nitrogen supply adequacy, providing reliable technical support for nitrogen status diagnosis and precision nitrogen management in sorghum cultivation.