Valid incidence rates of uveal melanoma (UM) from German population-based cancer registries are currently not available due to under-reporting. We conducted two case-control studies on UM at a reference centre for eye tumours and show the influence on population-based incidences of UM when data from case-control studies are linked with a cancer registry. The first case-control study (1996-1998) recruited 13 UM cases aged 35-74 years and the second case-control study (2002-2003) recruited 20 UM cases aged 20-74 residing within the population covered by the Münster Cancer Registry. After record linkage, age-truncated and standardized (World Standard Population) incidences with and without the record linkage were compared. Incidence rates based on routine cancer registration increased by a factor of 1.7 (1996-1998, age group 35-74 years) and 3.7 (2002-2003, age group 20-74 years) after record linkage with the case-control data. The supplemented age-standardized incidence of UM is 8.6 per million (20-74 years, 2002-2003) compared with the unsupplemented incidence of 2.3 per million. UM unknown to the registry were less often morphologically verified than those known to the registry. Cancer registries relying on pathology reports underestimate UM incidences if eye-preserving treatments are introduced. Close co-operation between cancer referral centres and cancer registries can substantially improve the completeness of registration.