We report 3 years of experience with quarterly assessments of the self-reported health of dialysis outpatients using the Short Form-36 (SF-36) Health Survey. The SF-36 measures eight different dimensions of health: physical function, role limitations due to physical problems, bodily pain, general health perceptions, vitality, social function, role limitations due to emotional problems, and general mental health. On each dimensions, the respondent receives a score from 0 to 100. A higher score indicates better health. Program logistics and results are described, including reliability coefficients, standard deviations, and standard errors of measurement for the SF-36 in this patient population. The SF-36 was completed an average of 4.4 times by 112 patients over an average 14-month follow-up; 40 patients responded at least six times. On most dimensions, the dialysis patients' scores were lower and more variable than those of a sample of the general US population matched for age and sex. For example, mean initial physical function score (+/- SD) was 48.5 +/- 31.2 for the dialysis patients and 84.8 +/- 23.3 for the general population. The mean initial general health perception score of the dialysis patients was 43.7 +/- 23.9 with 71.9 +/- 20.3 for the general population. In contrast, general mental health scores were more comparable. The mean initial general mental health score was 69.6 +/- 17.5 for our patients and 75.5 +/- 18.0 for the general population. Reliability (Cronbach's alpha) ranged from 0.77 for general mental health to 0.93 for physical function. The 95% confidence intervals around individual scores were approximately 20 points, except for role-physical and role-emotional, which were both approximately 30 points. Two case reports compare information obtained from the SF-36 with the dialysis team's assessments of the patient, as recorded in the medical record. The two patients' comments on reviewing their SF-36 results are also summarized. Patient reactions to the health status assessment program are explored, and potential benefits and areas for further work are outlined. Serial measurement of the health status of dialysis patients allows the recognition of clear patterns in individual patient responses. These patterns sometimes suggest that the patient is either substantially more or less impaired than the dialysis team had thought. Changes in these patterns, both transient and protracted, frequently exceed 95% confidence intervals for patient-level scores. Longitudinal health status profiles of individual dialysis patients, accumulated by repeated self-assessment using the SF-36, enhance rather than merely restate the providers' intuitive judgments expressed in the medical record.