BACKGROUND:The dialysis practice guideline in Japan sets a limit on the allowed interdialytic weight gain (IDWG) <6%. However, the effects of relative gain of fluid volume to body weight may differ in presence of morbid conditions. Here, we examined whether or not the associations between IDWG and mortality differ by serum albumin (sAlb), a nutritional and illness marker. DESIGN:The study type used was prospective cohort study. SUBJECTS:Patients who participated in the Japan Dialysis Outcomes and Practice Pattern Study (phase 1-4 [1999-2011]) and received thrice-weekly hemodialysis. METHODS:IDWG was the exposure of interest and was collected every 4 months, divided into 7 categories as follows: <2%, 2% to 3%, 3% to 4% (reference), 4% to 5%, 5% to 6%, 6% to 7%, and >7%. sAlb was treated as both an effect modifier and confounder and dichotomized into ≥3.8 g/dL and <3.8 g/dL segments, according to the protein-energy wasting criteria proposed by the International Society of Renal Nutrition and Metabolism. MAIN OUTCOME MEASURE:The outcome in this study was all-cause mortality. RESULTS:A total of 8,661 patients were analyzed. Time-varying Cox regression analyses revealed that, when sAlb was ≥3.8 g/dL, an IDWG >7% was associated with greater risk of mortality (adjusted hazard ratio [AHR] 2.74; 95% confidence interval [CI], 1.49-5.05). When sAlb was <3.8 g/dL, however, IDWGs <2% (AHR 1.89; 95% CI, 1.50-2.39) and 4% to 5% (AHR 0.75; 95% CI, 0.58-0.96) were associated with mortality (P for interaction = .001). Cubic spline analyses showed that the mortality increased when IDWG exceeded 6% for patients with sAlb ≥3.8 g/dL; in contrast, for patients with sAlb <3.8 g/dL, the mortality increased when IDWG was <3% and decreased when IDWG was between 4% and 6%. LIMITATION:The main limitation was possible residual confounding. CONCLUSIONS:The direction and magnitude of the associations between IDWG and mortality were modified by sAlb. Dialysis experts should take these results into account when revising the clinical practice guidelines.