Pine wilt disease has severely damaged pine forest resources in China, emphasizing the need for early and accurate diagnostics to prevent, control, and ensure national forest ecological security. Current diagnostic techniques include forest symptom diagnosis, pathogenic nematode identification, and the flow glue method, but these techniques have limitations in diagnosing needles before or at the stage of very few discolorations. Hence, a needle resistivity detection method based on spectral analysis of pine wilt disease is proposed. The study collected coniferous reflection spectral data of Pinus massoniana (8～9 years old) inoculated with Bursaphelenchus xylophilus in the wild, measured at different times using the Ocean Optics USB2000+. The average spectral reflectance at the canopys upper, middle, and lower positions was taken as the spectral reflectance of the plant. The needle cross-section was approximated to an ellipse, and a 4 cm section was cut from the middle of the needle to measuring its width, thickness, and resistance value using a M4070 LCR tester to calculate the resistivity. The original spectrum (OR) underwent spectral transformations using the first derivative (FD), second derivative (SD), logarithmic transformation (LOG), reciprocal transformation (1/R), and continuum removal (CR) methods. Characteristic bands were extracted from the original spectrum and each transformed spectral data using the random forest algorithm to invert the needle resistivity. The least squares support vector machine (LSSVM) algorithm analyzed the modeling effect of selected feature bands and the needle resistivity, identifying the best prediction model of needle resistivity. The study found that the needle resistivity of P. massoniana inoculated with B. xylophilus and the control reached a significant difference (p<0.01) in the early stages after a very small number of coniferous discolorations. The comprehensive performance of the spectral data shows that the secondary derivative transformation was found to be the best, with the characteristic bands being 594.986, 646.107, 646.451, 782.896, 784.841, 839.164, 863.890, 902.021, 947.901, and 962.315 nm. The study established that the prediction model established by SD-RF-LSSVM showed the highest accuracy, with an average R2 of 0.848 and an MAE of 32.331 and 7.067 for the modeling set and verification set, respectively. Compared to the model established using raw data (OR), this models R2 increased by 4%, and MAE decreased by 2.5% and 18.9%, demonstrating the feasibility of inverting the needle resistivity using the needle reflectance spectrum. Overall, this study provides a rapid estimation method for needle resistivity and offers ideas and methods for early diagnosis and monitoring of pine wilt disease based on remote sensing.