The fate of octogenarians reaching end-stage renal disease (ESRD) is poorly defined, and implicit dialysis rationing may be practiced in this age group. The main objectives of this study were to analyze the characteristics of pre-ESRD octogenarians offered dialysis or not and to identify factors influencing mortality while on dialysis, to improve prognosis assessment and decision-making. In this single-center cohort, 146 consecutive pre-ESRD octogenarians were referred to a nephrology unit over a 12-yr period (1989 to 2000). Main outcome measures were baseline characteristics of patients offered dialysis and conservative therapy and overall and 1-yr survival according to effective treatment. A therapeutic decision was made for 144 patients. Octogenarians who were not proposed dialysis (n = 37) differed from those who were proposed dialysis (n = 107) mainly in terms of social isolation (43.3% versus 14.7%; P = 0.03), late nephrologic referral (51.4% versus 28.9%; P = 0.01), Karnofsky score (55 +/- 18 versus 63 +/- 20; P = 0.03), and diabetic status (22.2% versus 6.5%, P = 0.008). Six patients refused the dialysis proposal. During the 12-yr observation period, 99 patients died (68.7%). Median survival was 28.9 mo (95% CI, 24 to 38) in patients undergoing dialysis, compared with 8.9 mo (95% CI, 4 to 10) in patients treated conservatively (P < 0.0001). In multivariable piecewise Cox analysis, independent predictors of death within 1 yr on dialysis were poor nutritional status, late referral, and functional dependence. Included in a survivor function, these covariates predict groups with low and high 1-yr mortality risk. Beyond 1 yr on dialysis, the only independent predictor of death was the presence of peripheral vascular disease. It is concluded that beside a patient's individual refusal, late referral, social isolation, low functional capacity, and diabetes may have oriented medical decision toward withholding dialysis in a significant proportion of pre-ESRD octogenarians. Although most patients on dialysis experienced a substantial prolongation of life, identification of mortality predictors in this age group should improve the process of decision-making regarding the expected benefit of renal replacement therapy.