The goal of this study was to develop an experimental-based online near-infrared (NIR) spectroscopic sensing system for the quality measurement of three major indicators of cow milk (fat, protein, and lactose), solids not fat (SNF) and somatic cell count (SCC) for each cow during milking. Milk samples were obtained from twenty-four Holstein cows. Near-infrared spectra of raw milk were obtained every 20 seconds during milking with an automatic milking system with wavelengths ranging from 700 to 1050 nm using NIRS sensing system. The three major milk quality indicators and SNF were measured using MilkoScan instrument, and SCC was measured using Fossomatic instrument. Calibration models were created using partial least squares regression analysis to validate the models’ precision and accuracy. The milk quality indicators for each milking time were determined by comparing the average predicted milk value and milk yield obtained every 20 seconds with the reference value for one milking time. The results obtained showed that the measurement accuracy for every 20 seconds during milking and at one milking time was very good and comparable. These results demonstrated that the NIR spectroscopic sensing system developed in this study could be used for online milk quality measurement in an automated milking system. The system could provide dairy farmers with information about the milk quality and health condition of each cow during milking, thereby providing them with feedback to improve dairy farm management. By using this NIRS sensing system, dairy farmers would be able to overcome the problem of individual cow management, produce high-quality milk, and achieve precision dairy farming.