Swedish population-based incidence and mortality rates for cancer of the uterine cervix, both in situ and invasive, during the period 1958 to 1981 were determined by means of a dynamic model. This new approach describes without any preconceptions the development of the disease as a sequential process over the stages cancer in situ, invasive cancer before and after diagnosis, and death. The strong disturbance of the steady-state situation that occurred after the introduction of cytological mass screening in the early 1960s permitted the use of a computerized identification technique. The whole natural history of cervical cancer could thus be identified and described consistently, with the mutual compatibility between statistical data, structure, parameters, and the states and flows between the states. The estimated age-specific incidence of cancer in situ increased rapidly to a maximum of 650 per 10(5) woman-years at the age of 30 years, after which it declined, and that of invasive cancer to a maximum of 55 per 10(5) at the age of 43. The natural history of cervical neoplasia did not differ appreciably between eight successive 5-year birth cohorts. The proportion of cases of new cancer in situ that progressed to invasive cancer was 12.2%, with a mean duration of the in situ stage in these cases of 13.3 years. The preclinical phase of the invasive stage (without screening) lasted on average about 4 years.