Abstract: In recent decades, heavy metal pollution in farmland has become more and more serious in our country, which exposes great detrimental effects on crop production and people's health. It's essentially necessary to monitor heavy metal in farmland soil qualitatively and quantitatively. Hyperspectral remote sensing has many characteristics, such as multiband, continuity and high resolution. It provides a new way for understanding the content of heavy metal in farmland soil rapidly and accurately. This paper focused on mechanism, impact factor, spectral characteristics and prediction model of hyperspectral remote sensing. The current limitations and expectations of hyperspectral remote sensing technology were also proposed.