Uveal melanoma (UM) is the most common intraocular malignancy in adults. Approximately 40-50% of UM are fatal as a result of metastatic disease, usually occurring to the liver. Monosomy 3 and gains of chromosome 8q are the strongest genetic predictors of the metastatic disease. Initially, routine genetic tumour typing in liverpool was based on fluorescence in-situ hybridisation (FISH). However, like others, our outcome analyses demonstrated that FISH has low sensitivity. Therefore, we selected Multiplex ligation-dependent Probe Amplification (MLPA) for UM prognostication in our routine practice. The aim of the Chapter's 2 project was to evaluate MLPA's sensitivity in the detection of chromosomal abnormalities in UM and its value of predicting metastatic disease using archival frozen UM material. We were able to confirm that metastatic death correlated most strongly with unequivocal chromosome 3 losses and gains on 8q. We also found that poor prognosis also correlated with equivocal/borderline MLPA values. We hypothesized that the cause of these equivocal MLPA results was intratumour heterogeneity with disomy 3 cell populations diluting monosomy 3 clones. We investigated this hypothesis in Chapter's 3 project using MLPA on formalin-fixed paraffin embedded (FFPE) material and we found that 75% of the UMs examined in our cohort were indeed heterogeneous for 1-26 of the 31 loci on chromosomes 1p, 3,6 and 8 tested by MLPA. While applying MLPA to the FFPE tumour material, we observed that the MLPA data quality was reduced compared to the frozen tumour material examined in Chapter 2. These were, however, two completely different tumour cohorts, so the results could not be compared directly. Therefore, in Chapter 4, we compared MLPA data obtained from adjacent freshly sampled frozen and fixed UM specimens, in order to investigate the effect of the formalin fixation on DNA quality and on MLPA data. We concluded that FFPE material may be as informative as frozen material; however, special rules and caution have to be applied in the interpretation of some results. Having collected MLPA data to 480 UMs, we also aimed to identify the common aberrant chromosomal pathways during UM tumorigenesis. In Chapter's 5 project, we found that UM may develop through more than two modes of progression and we observed that several chromosomal combinations are likely to be present in some pathways but not in the others. These findings are clinically relevant, as the knowledge on the possible combinations of chromosomal abnormalities in UM may help interpret MLPA data obtained from poorer quality FFPE material. Finally, in Chapter 6 and 7 we developed two algorithms, which improved MLPA data analysis and data selection, respectively. These algorithms were implemented in either MatLab or Microsoft Excel, which significantly reduced the time and complexity of MLPA data analysis. In conclusion, MLPA is a reliable and economic molecular biological method, which can be applied to frozen and FFPE UM material for prognostication. Its application in dividing UM patients into low- and high-risk groups for metastasis is enhanced by the incorporation of both clinical and histomorphological parameters. We have outlined proposals for the improvement of the current UM MLPA kit.