The effect of fluid overload and variation on residual renal function (RRF) in peritoneal dialysis (PD) patients is controversial. Retrospective cohort study was designed. One-hundred and ninety PD patients with measured glomerular filtration rate (mGFR) ≧ 3 ml/min/1.73 m2 were recruit. Fluid status of every participant was assessed by bioelectrical impedance analysis (BIA) every 3 months for 1 year. The cohort was divided into three hydration groups, namely persistent overhydration (PO) group, intermittent overhydration (IO) group and normal hydration (NH) group. Additionally, participants were also divided into high or low fluid variation groups. The decline rate of RRF and the event of anuria were followed up for 1 year. The association of fluid overload with RRF loss was evaluated by Cox proportional hazard models adjusted for confounders. Thirty-six (18.9%) patients developed anuria. The decline rate of mGFR in both PO and IO groups were significantly faster than that of NH group (PO vs NH: -0.2 vs -0.1 ml/min/1.73 m2/month, p < 0.01; IO vs NH: -0.2 vs -0.1 ml/min/1.73 m2/month, p < 0.01). Kaplan-Meier analysis showed poorer RRF outcome in both PO and IO groups compared with that of NH group (PO vs NH: p < 0.001; IO vs NH: p = 0.006). Patients with high fluid variation had worse RRF survival than those with low fluid variation (p = 0.04). Adjusted Cox regression models indicated the hazard ratio of RRF loss in PO group was 8.90-folds higher (95% confidence interval 3.07-31.89) than that in NH group. These findings suggested fluid overload was independently associated with the decline of RRF in PD patients.