Analysis of long-term dialysis results is the cornerstone of renal replacement therapy evaluation. Elderly patients may be considered a crucial cohort, since subtle differences may be enhanced in a population of lower life expectancy. The aim of the study was an analysis, from the Piedmont Registry of Dialysis and Transplantation, of the results obtained in 1981-1992 (northern Italy, about 4,400,000 inhabitants, 21 dialysis centres, open acceptance since mid-1970s) in patients aged > or = 65 years (475 patients started treatment in 1981-1985, 1026 in 1986-1992). As a first treatment, during the 12 years considered acetate haemodialysis decreased sharply; bicarbonate haemodialysis is currently the standard treatment (68%). Peritoneal dialysis is stable (21%), and haemodiafiltration is increasing (8%). Shifts between treatments are frequent: 15% of elderly patients changed treatment at least once in 1991-1992. Nephroangiosclerosis/ischaemic renal disease, undefined causes and diabetes mellitus are the major causes of end-stage renal disease; 57.3% of patients have high risk conditions in addition to age. In this cohort of patients, mean age of new cases starting dialysis significantly increased in 1986-1992 (72.7 +/- 5.4 years) versus 1981-1985 (71.3 +/- 4.5; P < 0.001). Despite this, survival at 2 years increased significantly from 54.6% in the period 1981-1985 to 59% in the period 1986-1992 (P < 0.05). Even in an ageing dialysis population, therefore, choice of an open dialysis system with easy changes among treatments allowed improvement of survival results; further technical advances may help in maintaining present trends.