Objectives: To study factors influencing the recommendation for palliative (non-dialytic) treatment in patients approaching end-stage renal failure and to study the subsequent outcome in patients choosing not to dialyse. Design: Cohort study of patients approaching end-stage renal failure who underwent multidisciplinary assessment and counselling about treatment options. Recruitment was over 54 months, and follow-up ranged from 3 to 57 months. Groups were defined on the basis of the therapy option recommended (palliative or renal replacement therapy). Setting: Renal unit in a district general hospital serving a population of about 1.15 million people. Subjects: 321 patients, mean age ± SD 61.5 ± 15.4 years (range: 16–92), 57% male, 30% diabetic. Main Outcome Measures: Survival, place of death (hospital or community). Results: Renal replacement therapy was recommended in 258 patients and palliative therapy in 63 (19.6%). By logistic regression analysis, patients recommended for palliative therapy were more functionally impaired (modified Karnofsky scale), older and more likely to have diabetes. The comorbidity severity score was not an independent predictor. Thirty-four patients eventually died during palliative treatment, 26 of whom died of renal failure. Ten patients recommended for palliative treatment opted for and were treated by dialysis. Median survival after dialysis initiation in these patients (8.3 months) was not significantly longer than survival beyond the putative date of dialysis initiation in palliatively treated patients (6.3 months). 65% of deaths occurring in dialysed patients took place in hospital compared with 27% in palliatively treated patients (p = 0.001). Conclusions: In high-risk, highly dependent patients with renal failure, the decision to dialyse or not has little impact on survival. Dialysis in such patients risks unnecessary medicalisation of death.