The objective was to discuss the evolution of the real incidence of breast cancer. Observed incidence as calculated by cancer registries differs from the real incidence because of the artifacts brought by diagnostic procedures and case collection. Age-period-cohort models were applied to nearly 11,200 incident breast cancers collected by the Cancer Registry of Is? from 1983 to 2002 in women aged 30-84. We took into account prior knowledge and assumptions concerning the evolution of real incidence, diagnostic procedures, and collection of cases. In the age group 30-49, no real incidence increase was seen if we assume that diagnostic procedures and collection of cases were not impaired. In women aged 50-69, an increase of real incidence and intensive screening could explain the increase of observed incidence but exact quantifications are difficult. At most, the increase due to screening would reach 50%. In women aged 70-84, no real incidence increase was suspected if we assume that changes in clinical practices and screening led to more breast cancer cases collected.