An investigation has been performed on the scattering properties of cow's milk. The main aim of the study has been to determine the usefulness of photon migration techniques in the measurement of milk components, and the extent to which they can be applied. Online applications are also investigated. The milk components that influence the optical scattering properties the most are the fat content and the protein content. Fat in milk exists in the form of small globules in a range of diameters from under 1 μrn to as large as 20 μrn. These globules are the major contributor to the light scattering properties of the milk. Milk protein is also a contributor to the scattering, but to a much lesser extent. Another component of milk that is of interest is the somatic cell concentration. These cells are usually present in milk at a low level, ranging from about 5 x 104 cells mL- 1up to 2 x 105 cells mL-1• Their numbers in milk dramatically increase with the onset of a mastitis infection. The somatic cell count is one of the primary measurements used in the dairy industry to monitor udder health and mastitis. Since milk is a highly scattering medium, one needs to choose a suitable mathematical model to characterise it so that quantitative measurements can be made. The diffusion approximation, a simplified version of transport theory, was selected. The derivation and the limitations of the diffusion approximation were discussed and noted in order to ensure that any subsequent measurements made with milk fell within the limits of the theory. A photon migration instrument (PMI) was built using the principles from the diffusion approximation theory. The instrument was configured to work with multiple-distance measurements. It was tested with a suspension of uniform microspheres and was shown to return results that agreed with Mie theory. Experiments were performed using the photon migration instrument with homogenised and raw milk. A good correlation was found between the fat content and the reduced-scattering coefficient. Homogenised milk samples of varying fat content were made by mixing trim milk with homogenised cream while raw milk of varying fat content was obtained from a robotic milking system. Both milks were tested with3the photon migration instrument - the homogenised milk was tested off-line, while the raw milk was tested with the instrument beside the robotic milking system. It was found that the instrument could determine the fat content in homogenised and raw milk with a precision of ±1.0 % and ±2.0 % milk fat content respectively. Protein and somatic cell count had a poor correlation with either the absorption coefficient or the reduced-scattering coefficient at a 670 nm wavelength. The precision of the raw milk fat content measurements can be increased by separating the milk samples by breed. Grouping milk samples from Friesian cows separately from milk samples from Jersey cows yielded precisions of± 1.6 % and ± 1.4% respectively. Further increases in precision could be obtained by separating out the results per cow, resulting in fat percentage measurements with a precision as high as±0.8 %. One of the more significant conclusions is that the fat globule particle size distribution is similar within the breed of cow and certainly similar from day-to-day with a single cow. Fluorescence photon diffusion techniques using the DNA stain ethidium bromide were used to determine somatic cell count. The results of this work showed that high cell-count samples could be quantified, with a lower detection limit of 4 x 105 cells mL- 1• The lower detection limit was too high, and during attempts to reduce it, evidence for an ethidium bromide inhibitor was found. This inhibitor appeared to use up free ethidium bromide without changing its fluorescence lifetime. This was shown by experiments where the milk was diluted with Intra Lipid to remove the contribution that differences in the absorption and reduced-scattering coefficient would make. The discovery of the ethidium bromide inhibitor was another significant conclusion of this work, but further investigation of the inhibitor was beyond the scope of this dissertation. The primary conclusion of this thesis is that photon migration techniques can measure some milk components online directly, notably fat content and somatic cell count. Improvements to the instrument are required however, in order to improve the precision of the measurements to the level required by the dairy industry.