Abstract: The contents of organic matter, related elements and heavy metals in sediments are the basis for the study of paleoenvironment and ecological environment evaluation. The near-infrared (NIR) diffuse reflection spectroscopy technique is a simple, rapid, and low cost method to determine the components in sediments. In this article we overview the standard procedure for sampling, the method of sample preparation and mathematical modelling for the chemical composition analysis in sediments. Furthermore, we introduce the selection of modelling, spectrum pretreatment and regression analysis to improve the accuracy of the quantitative analysis for a near-infrared spectrum model. The NIR diffuse reflection technique has been advocated to measure the contents of total organic carbon, total nitrogen, total phosphorus, biogenic silica, heavy metals etc. However, there is still plenty of work to be done since the application in sediments is not well established. At present, further study need to be carried out for the mechanism of measuring error in sediments. Properly chosen sample collection after suitable spectrum-pretreatment with optimization of mathematical models should improve the precision and accuracy of measurements.