The quality of life of patients with end-stage renal disease was estimated using the time trade-off technique. The sample included 103 transplant, 60 hospital hemodialysis, 57 home/self-care, and 52 continuous ambulatory peritoneal dialysis patients. Test-retest reliability was high: intra-class correlation coefficient 0.81 (p less than 0.001). The correlations of the time trade-off with the Spitzer Quality of Life index and a visual analogue scale completed by the nephrologists, nurses, friends/relatives, and the patient were positive and statistically significant, but still relatively low (r = 0.22-0.43; p less than 0.01). The time trade-off demonstrated evidence for discriminative construct validity by ordering treatment groups according to a priori prediction. The mean time trade-off values and standard deviations (where death is 0 and full health is 1) were 0.43 (0.26) for hospital hemodialysis 0.49 (0.23) for home/self-care hemodialysis, 0.56 (0.29) for continuous ambulatory peritoneal dialysis, and 0.84 (0.24) for transplant. Analysis of variance showed transplant to be different from all other groups (p less than 0.001) with age, sex, time with end-stage renal disease, and work status making no significant independent contribution. The partial correlation coefficients between time trade-off score and items in the physical, social, and emotional functioning sub-scales of the Rand questionnaire showed that physical functioning was far more important than social or emotional functioning. The time trade-off is reliable, demonstrates evidence for validity, and suggests that the quality of life for patients with end-stage renal disease is much poorer than that reported previously.