The aim of this study was to determine the incidence, prevalence, survival and selective demographic characteristics of scleroderma occurring in South Australia over the 10-year period 1993-2002. Analysis of the database of the South Australian Scleroderma Register: a population-based register established in 1993. Patients with scleroderma resident in South Australia (n = 353 at 2002) were ascertained from multiple sources and clinical and demographic data were obtained from mailed questionnaire and from review of computerized hospital databases, case notes or referring letters. Time-space cluster analysis was carried out according to the Knox method. Control data were obtained from the Australian Bureau of Statistics census. The mean prevalence was 21.4 per 10(5) (95% confidence interval 20.2-22.6) and the mean cumulative incidence of 1.5 per 10(5) (95% confidence interval 1.32-1.73) with no significant change in incidence over the study period (P = 0.13). Cumulative survival improved over the study period, with patients with diffuse disease having significantly reduced survival (as compared with limited disease, P < 0.001). The proportion with diffuse disease ( approximately 22%) remained steady. There was a small but significant predisposition in patients with a continental European birthplace (P < 0.001). A family history of scleroderma was noted in 1.6% with lambda1 (familial risk) of 14.3 (95% confidence interval 5.9-34.5). However, a family history of systemic autoimmunity (especially rheumatoid arthritis) was more common (6%). No socioeconomic stratification, temporal clustering nor spatio-temporal clustering was observed either at time of initial symptom or at 10 years before disease onset. Scleroderma occurs relatively infrequently in South Australia with no significant change in incidence observed over the 10-year study period. However, cumulative survival has improved. Identified risk factors include family history of scleroderma (risk approximately 14-fold), female sex (risk approximately 5-fold) and European birthplace (risk approximately 2.5-fold); however, the majority of the disease variance appears unexplained. A stochastic explanation based on genetic instability is favoured to explain this paradox.