Abstract: Presently, most oilfields in China have reached their middle and late years, and there is a high percentage of water in the crude oil in these oilfields. Accurate measurement of water content in crude oil is challenging, but of great significance to the adjustment of water injection strategy, the evaluation of crude oil production capacity, and the prediction of oil well lifespans. Current detection methods, however, all have significant detection errors and defects, which urgently call for new methods and equipment to improve detection efficiency. The components of crude oil produced liquid absorb light signals of different wavelengths differently, and near-infrared (NIR) light source achieves high penetrability. Employing a 1 500 nm light source, this paper designs a device that can quickly detect the water holdup rate of oilfield produced liquid and thus can be used in high water holdup conditions. The system signal is linearly processed based on Lambert Beer’s law for water retention and absorbance data, achieving accurate measurement of water holdup detection of high water content crude oil. This rapid detection system effectively measures the water holdup of the crude oil, with a measurement range of 70% to 100% for the water holdup of the crude oil produced liquid. When the water holdup is over 70%, the testing error stands at 1.5%. In addition, the impact of actual underground factors such as flow rate, temperature, salinity, and sediment concentration on water holdup is studied, and relevant correction factors are summarized. This study provides a solid theoretical and practical foundation for the underground operation of a novel fast infrared light source fiber optic detection system.