Clinical impact of changes of renal function (RF) in heart failure (HF) hospitalization is controversial. This study aimed to clarify whether clinical impact of changes of RF during HF hospitalization depends on the intrinsic RF. In 786 hospitalized HF patients, RF were classified into 3 grades based on estimated glomerular filtration rate (eGFR, mL/min/1.73 m²) at discharge; ≥ 60 (n = 243), < 60 and ≥ 30 (n = 400), and < 30 (n = 143). Increase and decrease of serum creatinine over 0.3 mg/dL during HF hospitalization were defined as worsening renal function (WRF) and improved renal function (IRF), respectively, and remaining subjects were defined as stable RF. The primary endpoint was a composite of cardiovascular death and rehospitalization for HF. In all patients, WRF was not associated with clinical outcomes, although eGFR has a significant association with prognosis. Clinical outcomes did not differ between changes of RF patterns in both preserved and severely impaired RF groups. In contrast, IRF, not WRF, was an independent predictor of clinical outcomes in the moderately impaired RF group (HR 1.965, 95% CI 1.09–3.18, p = 0.01). Only in patients with moderately impaired RF, changes of RF were associated with clinical outcome, and IRF was an independent predictor of clinical outcomes.