A tool to predict peritonitis-associated treatment failure among peritoneal dialysis (PD) patients has not yet been established. We conducted a multicentre, retrospective cohort study among 1,025 PD patients between 2006 and 2016 in Thailand to develop and internally validate such a tool. Treatment failure was defined as either a requirement for catheter removal, a switch to haemodialysis, or peritonitis-associated mortality. Prediction model performances were analysed using discrimination (C-statistics) and calibration (Hosmer-Lemeshow test) tests. Predictors were weighted to calculate a risk score. In total, 435 patients with 855 episodes of peritonitis were identified; 215 (25.2%) episodes resulted in treatment failure. A total risk score of 11.5 was developed including, diabetes, systolic blood pressure <90 mmHg, and dialysate leukocyte count >1,000/mm3 and >100/mm3 on days 3-4 and day 5, respectively. The discrimination (C-statistic = 0.92; 95%CI, 0.89-0.94) and calibration (P > 0.05) indicated an excellent performance. No significant difference was observed in the internal validation cohort. The rate of treatment failure in the different groups was 3.0% (low-risk, <1.5 points), 54.4% (moderate-risk, 1.5-9 points), and 89.5% (high-risk, >9 points). A simplified risk-scoring scheme to predict treatment failure may be useful for clinical decision making regarding PD patients with peritonitis. External validation studies are needed.