Scleroderma is a multisystem disease which in many populations is reported to have a substantially reduced survival. One problem with many of the published studies is the inclusion of patients with differing lengths of disease at baseline, with the possibility of left censorship bias. The aim was to determine the survival and its predictors in a cohort of scleroderma patients with new onset of disease. All 283 new patients referred to one of the authors (CMB) who had a reported disease onset after 1 January 1982 were studied. Detailed clinical and laboratory information at baseline and at 1 yr were extracted from the structured records. Vital status as at 1 December 1993 was determined using the UK's NHS Central Register. Expected numbers of deaths were obtained by applying local age and sex all-cause mortality rates. In total, there were 1871 person-years of follow-up with crude mortality rates of 3.9 and 2.6%/year, respectively, in males and females. The mortality rate remained approximately constant over the first 10 yr of follow-up. In comparison with expected survival, based on the age distribution of the cohort, there was an overall 4-fold increased mortality rate in this group which was higher in females [standardized mortality ratio (SMR) 4.6, 95% CI 3.2-6.2] than in males (SMR 3.2, 95% CI 1.8-5.0). Analysis of cause-specific mortality showed that 79% of the excess deaths were scleroderma related, but this proportion was substantially lower in males than in females (67% vs 83.9%). These data confirm the increased mortality rate associated with scleroderma, but these rates are lower than those reported from other centres. Females have a higher age-adjusted mortality which is reflected in the different pattern of mortality between the sexes.