In recent times, there has been an urgent requirement for a technique that can be used by dairy farmers to assess the quality of milk of an individual cow during milking. A near-infrared (NIR) spectroscopic sensing system was designed on an experimental basis for the online assessment of the three major milk constituents (fat, protein, and lactose), solids not fat (SNF), milk urea nitrogen (MUN), and somatic cell count (SCC). This system was used to obtain the NIR spectra of non-homogenized milk during milking over a wavelength range of 700 to 1,050 nm. Calibration models for predicting three major milk constituents, SNF, MUN and SCC of non-homogenized milk were developed, and the precision and accuracy of the models were validated. The coefficients of determination, standard errors of prediction, and bias values showed high levels of precision and accuracy for the prediction of the considered parameters. The results indicated that the developed NIR spectroscopic sensing system can be used to assess milk quality in real-time during milking. This system can provide dairy farmers with information concerning milk quality and physiological condition of each cow, and can therefore optimize dairy farm management.