Nuclear Magnetic Resonance was used to investigate selected fruit with internal defects and at different stages of ripening. This was motivated by the need for an industrial NMR sensor of fruit and vegetable quality capable of operating at typical industrial conveyor speeds. Such a sensor requires the development of NMR protocols capable of determining fruit and vegetable quality in a single-shot manner. It also requires knowledge of which combination of NMR parameters is most sensitive to the quality factor of interest. In this thesis we therefore address this issue by exploring the off-line relationship between NMR parameters and fruit quality. We mainly focus on three areas: the detection of mealiness in apples, the single-shot measurement of Brix, and a single-shot measurement of oil in avocado. One dimensional techniques, measuring transverse relaxation times, were used as starting point. The correlation between mealiness and one dimensional was not observed at low fields. Therefore two-dimensional approaches were taken. The first of these was correlation spectroscopy, which showed that the of the water associated with the cell wall in mealy apples is much longer that that of fresh apples. This in principle could be used to classify apples. The differences between fresh and mealy apples observed with the second two-dimensional protocol, could ~ot be used to classify apples. However, when the technique was applied to avocado, it became clear that the difference between the diffusion coefficients of water and oil could be exploited for an on-line application. A fast pulse sequence was developed and the off-line measurements showed a good correlation between oil content and the echo ratio. A similar protocol was developed for the on-line measurement of Brix value in apple and strawberry.