Current fuel management strategies for light water reactors (LWRs), in countries with high back-end costs, progressively extend the discharge burnup at the expense of increasing the 235U enrichment of the fresh UO2 fuel loaded. In this perspective, standard non-destructive assay (NDA) techniques, which are very attractive because they are fast, cheap, and preserve the fuel integrity, in contrast to destructive approaches, require further validation when burnup values become higher than 50 GWd/t. This doctoral work has been devoted to the development and optimisation of non-destructive assay (NDA) techniques based on gamma-ray emissions from irradiated fuel. It represents an important extension of the unique, high-burnup related database, generated in the framework of the LWR-PROTEUS Phase II experiments. A novel tomographic measurement station has been designed and developed for the investigation of irradiated fuel rod segments. A unique feature of the station is that it allows both gamma-ray transmission and emission computerised tomography to be performed on single fuel rods. Four burnt UO2 fuel rod segments of 400 mm length have been investigated, two with very high (52 GWd/t and 71 GWd/t) and two with ultra-high (91 GWd/t and 126 GWd/t) burnup. Several research areas have been addressed, as described below. The application of transmission tomography to spent fuel rods has been a major task, because of difficulties of implementation and the uniqueness of the experiments. The main achievements, in this context, have been the determination of fuel rod average material density (a linear relationship between density and burnup was established), fuel rod linear attenuation coefficient distribution (for use in emission tomography), and fuel rod material density distribution. The non-destructive technique of emission computerised tomography (CT) has been applied to the very high and ultra-high burnup fuel rod samples for determining their within-rod distributions of caesium and europium fission-product radionuclides. As indicated above, results provided by the transmission tomography measurements were employed in the emission tomography reconstruction phase, together with a calculated global efficiency matrix and input sinograms derived from the processing of measured projections. Different tomographic algorithms were tested and "tuned", on the basis of known test distributions, before being applied to the actual fuel rod measurements. Amongst the various possibilities, the Paraboloidal Surrogates Coordinate Ascent (PSCA) penalised likelihood method has been chosen for presentation of the final results, because it ensures high precision, especially in resolving the most difficult peripheral regions of the rods. The results of the emission tomography have indicated large central depressions in the caesium distributions, but of varying extent from sample to sample. Particularly interesting is the case of the 126 GWd/t sample, showing a very deep central depression (a factor of ∼2.5 for 137Cs, a factor of ∼3 for 134Cs). Differences in the relative activity distributions of 137Cs and 134Cs have, in fact, been observed for all the samples. The depression of 134Cs is more marked than that of 137Cs, probably due to the different origins of the two isotopes. In contrast, the europium shows an almost flat distribution. In order to support the tomographically measured caesium distributions, the results of destructive chemical techniques applied on samples from the same fuel rod (126 GWd/t sample) were examined and found to show reasonably good agreement with the tomography, thus confirming the depressed distributions at the centre of the rod. In addition to the tomographic reconstructions, the present research has also investigated the possibility to use single isotope activities, and/or isotopic concentration ratios from 134Cs, 137Cs, and 154Eu, as burnup indicators at very high and ultra-high burnups. The corresponding non-destructive measurements, performed using three different approaches, have been compared with chemical assays, as well as with reactor physics calculations (CASMO). Whereas the chemical results confirm the current gamma-spectroscopic measurements within uncertainties, the agreement between measurements and calculations is not satisfactory. It has been shown that certain indicators, well established for application at low and medium burnups, suffer a serious loss of reliability at high burnup. Nevertheless, the possibility of successfully employing other burnup monitors has been clearly highlighted. Considering that the overall effort required for the destructive chemical analysis of fuel samples is much greater, the current investigations have clearly demonstrated that non-destructive gamma spectrometry, in conjunction with corresponding transmission and emission tomography measurements, can indeed be considered a valid approach for characterising fuel burnup in the high/ultra-high range.