To provide a basis for the evaluation of mammographic screening programmes in Denmark, a study was undertaken of the regional differences in breast cancer incidence and mortality. All 16 regions were followed for the 20 year period, 1970-89, before the start of the first population-based mammographic screening programme in the Copenhagen municipality in 1991. Multiplicative Poisson models were used for the analysis. In general, the incidence increased during this period from 55 to 70 [per 100,000 standardised world standard population (WSP)], and the analysis shows this to be most pronounced among women below age 60. The mortality was more stable, changing only from 24 to 28 (per 100,000 standardised WSP), but a significant increase occurred in the late 1980s. The study showed regional differences in both incidence and mortality of breast cancer in Denmark. Both the incidence and the mortality varied between the regions, with maximum differences of 22%. The analysis showed no variation in the time trends in the different regions, and thus indicates that the use of a regional comparison group would be a valid basis for evaluation of the Copenhagen programme. Our study, however, underlies the difficulties inherent in the evaluation of screening programmes without internal control groups.