In patients receiving maintenance hemodialysis, laboratory indices (such as serum albumin concentration) are predominantly utilized to assess well-being, while measures of functional status are rarely applied. However, the serum albumin concentration declines with advancing age, and the mean age of patients starting maintenance hemodialysis is now over 63 years. Using a 14-level modified Karnofsky activity scale, we measured baseline functional status in 522 randomly selected hemodialysis patients and prospectively monitored them for 3 years to determine the predictive value of our modified Karnofsky score for mortality. At onset of study, serum albumin and creatinine concentrations as well as hematocrit were measured and the comorbid conditions documented. At baseline, the 522 subjects (270 women and 252 men) included 327 blacks (63%), 154 whites (29%), 31 Hispanics (6%), and 10 Asians (2%) of mean age 59 +/- (SD) 15 years. The mean duration of end-stage renal disease was 4 +/- 3.6 years, and the mean serum albumin concentration was 3.7 +/- 0.4 g/dl. 166 (32%) of the patients died during the observation period. Cox regression analysis revealed inverse relations between mortality and both our modified Karnofsky score (p = 0.0001) and serum albumin concentration (p = 0.001). The predictive value of a low modified Karnofsky score for mortality persisted after analysis of subjects stratified according to serum albumin concentration (< 4 g/dl, n = 382, p = 0.0001 vs. > or = 4 g/dl, n = 140, p = 0.008). With a modified Karnofsky score (< 70 vs. > or = 70), the relative risk of death during the 3-year follow-up period was 1.44 (95% confidence interval 1.236, 1.675; p < 0.0001). Forward stepwise Cox regression analysis showed that advanced age (p = 0.0005), white race (p = 0.0009), diabetes mellitus (p = 0.01), and a low serum albumin concentration (p = 0.003) were independently associated with an increased risk of mortality during follow-up after adjustment for other factors. A modified Karnofsky score (p = 0.14) did not predict survival in the Cox model when other independent variables were included. We conclude that in patients with end-stage renal disease sustained on maintenance hemodialysis, a poor functional status (measured on a modified Karnofsky activity scale) is associated with early mortality. Periodic measurement of modified Karnofsky score is a simple, low-cost, and reliable means of identifying patients on dialysis at risk for early death.