A previous study at our center used the Charlson Comorbidity Index (CCI) (an index of comorbidity that includes age) to predict outcomes in a mixed group of incident and prevalent dialysis patients. The purpose of this study was to examine the usefulness of the CCI as a predictor in incident peritoneal dialysis (PD) patients and to examine whether it was a better predictor than simply the number of comorbid conditions or other known predictors, such as age, albumin level, diabetes, and cardiovascular disease. Since 1990, we have collected prospectively comorbidity data at the start of PD. All patients with known comorbidity and serum albumin and who did not have a prior history of hemodialysis or transplant were included (n = 268). Time at risk began at day 1 of PD training. Cox proportional hazards best subset selection was used to screen models to predict patient survival. Candidate models were analyzed further for proportionality and other model assumptions. Univariate analysis showed that significant predictors of mortality were CCI (chi-square = 43.3, P < 0.0001), age (chi-square = 23.7, P < 0.0001), cardiac disease (chi-square = 19.9, P <0.0001), number of comorbid conditions (chi-square = 15.6, P < 0.0001), serum albumin at the start of dialysis (chi-square = 15.3, P = 0.0001), and diabetes (chi-square = 4, P < 0.05). In multivariate analysis, CCI alone was the best predictor. The addition of serum albumin did not improve the model significantly (chi-square = 51.86 versus 49.34). For every increase of 1 in the CCI score, the relative risk of death was 1.54 (95% confidence interval, 1.36 to 1.74). CCI alone scored at the start of PD is a strong predictor of patient survival in incident end-stage renal disease patients on PD. This simple-to-calculate index would be useful to adjust for confounding in future studies and in the adjustment of case mix if Medicare moves to a capitated payment system.