Abstract: Cross-linked polyethylene (XLPE) cable main insulation is highly susceptible to moisture in humid environments, leading to water treeing and insulation degradation. To enable rapid on-site diagnosis, this study proposes a handheld near-infrared spectroscopy method based on characteristic bands, calibrated using a synchronous gravimetric–spectral method. A dual-band ratio model was established with 1700 nmas the reference band and 1940 nmas the detection band, effectively minimizing thickness variation, baseline drift, and scattering effects. Spectral analysis shows that the 1940 nmband offers superior dynamic range and sensitivity retention over 1450 nm, making it more suitable for quantitative moisture detection. A least-squares calibration yielded a coefficient of determination (R²) of 0.989, confirming strong agreement between predicted and measured values. The portable device developed from this model was validated under laboratory and field conditions, with deviations from the oven-drying standard kept within 3%. These results demonstrate that the system provides rapid, robust, and quantitative detection of moisture inXLPEinsulation, supporting reliable cable operation and lifetime management.