Red earth sediments, widely distributed in southern China, are important archives of Quaternary environmental evolution in the mid-low latitudes. Hematite (Hm) within these sediments is a key indicator of dry and hot paleoclimates. However, traditional chemical methods fall short for rapid and precise quantification in complex soil matrices. Diffuse reflectance spectroscopy (DRS) combined with multiple linear regression (MLR) was used to quantify Hm content under complex mineral backgrounds. Applied to the Tongling red earth profile in the lower Yangtze River, this approach reconstructed the Hm content record since the Middle Pleistocene. The results show a good correspondence between the reconstructed hematite Hm content and the redness index (a*), validating the reliability of a* as a Hm proxy from an absolute data perspective. Moreover, this quantitative result provides evidence for deciphering the magnetic depletion mechanism in vermiculated red earth and demonstrates the utility of DRS in paleoenvironmental studies of southern Chinese red earth. Thus, DRS-based quantitative inversion is an effective tool for analyzing complex multi-source sediments and improves the ability to identify multi-stage pedogenic overprinting in southern red earth.