Whey has proven to be a rich source of nutrients that can bring added value to the consumer and increase the value of dairy streams. Beta-lactoglobulin and alpha-lactalbumin are the two main whey proteins that are found in milk and each have specific functional and nutritional characteristics that make them valuable ingredients in consumer foods and infant formula. Consequently, protein fractionation processes have been developed to separate the two target proteins and selectively enrich whey protein concentrates. The subject of this thesis is the monitoring of a protein fractionation process that uses membrane filtration to separate the whey proteins. Monitoring and control of filtration processes is currently done with discontinuous, off-line measurements based on expensive and slow wet chemistry methods. However, the development of continuous, high throughput processes in the industry necessitates faster and more efficient monitoring of quality attributes to achieve optimal process control. This project develops spectroscopic methods to quantify alpha-lactalbumin and beta-lactoglobulin in whey streams, providing real-time information about the performance of the fractionation system and in particular the membrane selectivity. After screening of several methods and calibration strategies, two spectroscopic techniques, near-infrared spectroscopy and ultraviolet/visual spectroscopy, have been tested on retentate and permeate streams, respectively, in Arla Food Ingredients’ processing facilities. Both methods have proven to be sufficiently sensitive for process monitoring in an at-line set-up, showing adequate precision and accuracy to detect key changes in the process. On-line ultraviolet/visual spectroscopy (UV-Vis) was tested in a laboratory set-up for real-time quantification of whey protein fractions in a permeate stream. The result showed that UV- Vis can be used to quantify changes in membrane selectivity during a filtration process. Near-infrared spectroscopy (NIRS) has been tested in a full-scale in-line set-up for investigating the feasibility of continuous selective protein quantification in a retentate stream. Although in-line analysis in full scale compromised spectral quality, protein quantification was still sufficiently accurate to describe key process behaviours across different production runs. Both the in-line and on-line set-up delivered acceptable results that show the applicability of both techniques for real-time selective protein quantification. The spectroscopic tools developed were coupled with chemometrics and process control methods to develop Process Analytical Technology (PAT) tools applicable in a manufacturing environment. The potential application of the developed tools for process understanding, optimisation and development are discussed in this work. A profitable protein fractionation process must be able to combine high selectivity and high productivity. Understanding, validating and improving a process requires an appropriate monitoring of the process performance, including of membrane selectivity. This project has proven that near-infrared (NIR) and ultraviolet/visual (UV-Vis) spectroscopy can be used as time and cost efficient analytical monitoring tools. The positive results obtained show the potential of both techniques being implemented within a process control strategy. In-line/on-line tools developed in this process elucidate some process kinetics hard to detect with at- line analysis and current monitoring of process (engineering) parameters. In conclusion, the results obtained in this project show the potential of spectroscopic tools and multivariate data analysis for implementation within a Quality by Design (QbD) strategy to optimise protein ingredients manufacturing.