Objective: Peritonitis is a severe complication in peritoneal dialysis (PD). This study was performed to identify predictors and establish a risk score for treatment failure in peritonitis patients.Methods: A single-center, retrospective observational study was conducted. The basic demographic characteristics, clinical and laboratory data of all patients with peritonitis during the study period were documented and analyzed. Multivariate logistic regression was applied to examine independent predictors of treatment failure, and a risk prediction score was established.Results: Three hundred fourteen episodes experienced by 241 patients were included in the final analysis. Logistic regression analysis indicated that PD duration (OR 1.017; P 0.005), fibrinogen (OR 1.327; P 0.021), high-density lipoprotein (OR 0.443; P 0.032), fungal infection (OR 63.413; P < 0.001), intestinal obstruction (OR 5.186, P 0.007), and diabetes mellitus (OR 2.451; P 0.018), hemodialysis history (OR 2.804, P 0.006) were independent predictors of treatment failure. The risk prediction score system showed a good calibration (P > 0.05) and discrimination (AUROC 0.80, P < 0.001).Conclusions: Fibrinogen, PD duration, fungal infection, hemodialysis history, concurrent intestinal obstruction, or diabetes mellitus were independent risk factors for a poor peritonitis outcome, while the high-density lipoprotein was a protective factor. This novel risk prediction score system may be used to predict a high risk of treatment failure effectively.