BACKGROUND Several factors have contributed to the substantial decline in mortality from cervical cancer registered in most areas of the world, i.e., improved sexual hygiene, changes in reproductive factors, cervical screening, and, possibly, improved treatment. Each of these components is evaluated through a systematic inspection of trends in incidence, mortality and survival rates registered for a well-defined population. METHODS Trends in incidence, mortality and survival from invasive cervical cancer over the period 1974-1991 were analysed for three separate age groups (< 55, 55-64, > or = 65 years), histological type and stage using data from the Cancer Registry of the Swiss Canton of Vaud. RESULTS Below age 55, the age-standardized (world standard) incidence rate was 9.3/100,000 women in 1974-76, it declined steadily thereafter down to 2.9 in 1986-88, but increased to 4.3 in 1989-91. In the age group 55-64, cervical cancer incidence remained around 40/100,000 to the end of the 1970s, but thereafter declined to 10.9 in 1989-91. No consistent trend was observed in elderly women, and the rate in 1989-91 (26.7/100,000) was similar to that in 1974-76 (33.7). The overall age-standardized cervical cancer incidence declined from 13.5/100,000 in 1974-76 to 5.8 in 1986-88, but rose to 6.4 in 1989-91. A similar pattern was observed for mortality, with a fall in rates in younger women between 1974 and 1985 (from 2.1 to 0.6/100,000), and a subsequent rise to 1.1/100,000 in 1989-91. A substantial decline in mortality was observed in women aged 55 to 64 since the early 1980's, from 17.2 in 1980-82 to 3.3 in 1989-91. No clear mortality trend was evident in older women. Overall, cervical cancer mortality declined from 4.3/100,000 in 1974-76 to 2.3 in 1989-91. The five-year relative survival rates were around 0.70-0.75 for younger women, around 0.60 for those aged 55 to 64, and 0.50 for elderly ones. In a Cox proportional hazard model, age and clinical stage were significantly related to survival, but not histological type and calendar period of diagnosis. No substantial change in survival from invasive cervical cancer was observed over the 18-year period considered, nor was there any notable change in the stage distribution over time. The proportion of adenocarcinomas, however, appears to have increased in the most recent calendar period. CONCLUSIONS These data reflect the impact of screening on cervical cancer rates, which, however, appeared restricted to women below age 65. An upward trend in cervical cancer incidence and mortality rates for younger women was also apparent, and there was no indication from these data of an improved survival for invasive cervical cancer patients over the last two decades. Extention of screening to elderly women appears to be a priority for reducing cervical cancer rates in this population.