A measure of comorbidity in dialysis patients must not only predict outcomes but also be reproducible and easy to obtain. Our primary purpose in the present study was to determine the inter-rater reliability of the Charlson comorbidity index (CCI) in peritoneal dialysis (PD) patients. Our secondary purpose was to evaluate the usefulness of annual rescoring of the CCI as a predictor of patient survival. We included in the study 100 consecutive patients (mean age: 52 +/- 16 years; 85% white; 39% with diabetes) who started PD between 1995 and 2000 at a single center: Two nurses independently scored the CCI at the start of PD. One nurse rescored each patient on the yearly anniversary of the start of PD. Patient survival was recorded for each year Kappa score and time-dependent analysis were applied. The kappa score between the two CCI scores at the start of dialysis was 0.93. (The average scores by the two nurses were 5.2 and 5.3.) Annual rescoring of the CCI demonstrated no increase in its predictive value regarding patient survival. However, given the minimal change in the CCI for the patient population in the present study, that question needs further study. Over time, the average CCI fell--an unsurprising result, because patients with higher CCI scores at the start of dialysis are the most likely to die. We conclude that the CCI is a reliable and easily applied tool for assessing comorbidity. Dialysis units should consider obtaining this measure at the start of dialysis in all patients. Repetitive annual scoring was not helpful in improving prediction of survival.