Phosphorus is an essential nutrient for plant growth and development. To further investigate the feasibility of hyperspectral data for estimating leaf phosphorus content of winter wheat throughout its entire growth cycle，this study conducted remote sensing monitoring of winter wheat in Qitai county，Xinjiang，in 2023. Hyperspectral data of winter wheat canopies were collected using an spectroradiometer and combined with laboratory-measured leaf phosphorus content. Correlation analysis was performed between leaf phosphorus content and both raw spectral data and spectral indices. Sensitive spectral bands and indices were used as input variables for stepwise regression，partial least squares regression，BP neural network regression，and random forest regression to predict leaf phosphorus content. The results demonstrated the successful construction of leaf phosphorus content estimation models applicable to winter wheat throughout its entire growth cycle in Qitai. Specifically，machine learning models，especially the random forest regression model，outperformed linear regression models at the jointing，flowering，and grain-filling stages，while the BP neural network regression model showed better performance at the booting stage. The coefficients of determination(R2)for the random forest regression model were 0.764，0.811，and 0.805 at the jointing，flowering，and grain-filling stages，respectively，with root mean square errors(RMSE)of 0.652，0.152，and 0.224. For the BP neural network regression model at the booting stage，the R2 was 0.772 with an RMSE of 0.313. These findings confirmed the effectiveness and feasibility of hyperspectral technology for estimating leaf phosphorus content in winter wheat，providing technical support for precision agricultural management and crop yield improvement. This study offered an application case for hyperspectral technology in crop phosphorus retrieval，which could help optimize phosphorus fertilizer application strategies and serves as a reference for future research.