The aim of this study was to analyse time trends, stage at diagnosis, survival and registration of population-based cohorts of breast cancer patients in selected Danish counties (in total 2504) in 1986 and 1996-1997. In 1986, no differences in the extent of disease were observed between the counties. Patients from one county (Funen) had centralised surgery, significantly more lymph nodes removed and a better survival in the multivariate analysis. In 1996-1997, mammographical screening had been implemented in Funen, leading to a significantly better stage distribution, whereas stage remained unchanged in the other counties. In Funen, survival was significantly better than in the other counties in univariate, but not in multivariate analysis. Survival increased significantly with time only in Funen. Inclusion in clinical trials increased over time and the coverage of the database in the Danish Breast Cancer Cooperative Group (DBCG) was high. However, patients not notified in DBCG had, beside older age, also worse stage of disease distribution and less extensive surgery. A difference in survival was observed between the counties. In 1986, this may be explained by a centralised surgical system in one county, whereas in 1996-1997 improvements could be due to an early diagnosis and other as yet unknown factors. The DBCG database cannot be considered as representative of the Danish population of breast cancer patients.