BACKGROUND--For uveal melanoma it has been demonstrated that aneuploidy correlates with worse clinical outcome. However, a striking variation in incidence of aneuploidy is reported for uveal melanomas. METHODS--Flow cytometry was used to study retrospectively DNA-ploidy of 132 uveal melanomas on paraffin embedded material. Thirty five patients received 2 x 4 Gy doses of irradiation 24 and 48 hours before enucleation. Correlation between DNA-ploidy and histopathological grading, largest tumour diameter, tumour height, tumour location, scleral invasion, and TNM classification was assessed. Survival analysis methods were used to investigate the predictive value of these variables on clinical outcome. RESULTS--Of the tumours 37% were aneuploid and 63% were diploid. Intratumour ploidy heterogeneity was minimal (92% concordance). A strong correlation (p = 0.009) was found between DNA-ploidy and cell type. No correlation was found between DNA-ploidy and other conventional prognostic variables. Irradiated melanomas were significantly more aneuploid than non-irradiated tumours (p < or = 0.01). CONCLUSION--In survival analysis DNA-ploidy and the largest tumour diameter were significant in predicting metastatic outcome (p < or = 0.03 and 0.01 respectively); histological cell type and tumour location were of borderline significance.