Hysterectomy fractions, or the inverse --the proportion (prevalence) of women with an intact uterus (and cervix) -are of interest for several reasons. 1 . Such fractions by age group are required to estimate proper population denominators for calculating incidence or mortality rates of cancers of the cervix and other pathological conditions of these tissues and medical procedures carried out on these structures (including hysterectomy itself), since women without a uterus (or cervix) are not at risk of such conditions or interventions. 2. With the advent of population-based cervical screening for pre-cancerous conditions using exfoliate cytology (Papanicolaou smears), it is necessary to estimate the population of women (by age) who should avail themselves of such testing for the purpclses of planning these services and estimating response rates.5 Furthermore, cervical screening rates in both whole populations5-* and those attending health services require denominators adjusted for hysterectomy fraction^.^ 3. Hysterectomy fractions are a way of displaying the cumulative consequences of hysterectomies in a p o p~l a t i o n . '~-~~ If hysterectomy (andor removal of the cervix) were rare occurrences, the population effect could well be ignored. However, removal of these tissues has occurred with such frequency that substantial proportions (20-30%) of middle-aged and older women in Australia no longer possess these parts of their reproductive s y~t e m .~. " -~~. '~ Furthermore, since removal of the uterus (and cervix) vary over time and between different sub-populations.,1617 hysterectomy fractions also exhibit secular and differential variation. Such differences cannot be ignored in calculations of disease rates and service planning.There are two approaches to the estimation of hysterectomy fractions. The empirical approach is to ask a representative sample of women at one point of time whether they have ever had their uterus or cervix removed; such a question was included in the Australian Health Surveys (AHS) of 1989-90 and 1995,I4.l5 and in other representative sample surveys."-13 This approach is constrained by: the representativeness of the sample, knowledge of women of the details of surgical procedures performed on them, small numbers available for analysis at subnational level, and whether the survey has been repeated sufficiently frequently (including during past decades) to provide secular information.The second approach is to construct cumulative probabilities of retaining a uterus (and cervix) by age group at particular periods in time from known prior hysterectomy rates of the cohorts of women which com: pose each ageThe problem with this approach is that available time series data on hysterectomy is often of insufficient length, necessitating back p r~j e c t i o n .~ In NSW, hysterectomy data can be obtained from the NSW Inpatient Statistics Collection (ISC). The earliest state-wide ISC hysterectomy data (partially based on a sample) are from 1979, and the series during the 1980s is incomplete.In th...