Real-world evidence describing the variation in serum creatinine (S-Cre) within 24 hours and its prognostic value is unknown. We enrolled 14 912 adults who received two S-Cre measurements within 24 hours at a tertiary hospital between 2003 and 2016. The study population was divided into four groups according to the hospital service settings where the baseline and second S-Cre were measured: Group 1, Outpatient-to-Outpatient; Group 2, Outpatient-to-ED (emergency department) or Inpatient; Group 3, ED-to-ED or Inpatient; and Group 4, Inpatient-to-Inpatient. The main predictors were the difference between the two S-Cre measurements (ΔS-Cre) and the percent change (ΔS-Cre%). The main outcomes were 30-day, 1-year, or 3-year all-cause mortality. A total of 6753 and 8159 patients with an increase and a decrease within-day ΔS-Cre, respectively. Among 6753 patients who had deteriorating ΔS-Cre or ΔS-Cre%, the adjusted hazard ratio (aHR) for 1-year all-cause mortality for each 0.1 mg/dL or 5% change in S-Cre was 1.09 (95% confidence interval [CI]: 1.07, 1.11) and 1.03 (95% CI: 1.03, 1.04). In 8159 patients with improving ΔS-Cre%, the aHR was 0.97 (95% CI: 0.94, 1.00). Groups 3 and 4 had statistically significant positive linear relationships between deteriorating ΔS-Cre% and 30-day and 3-year mortality. The optimal cut-offs for deteriorating ΔS-Cre% for predicting 30-day mortality were approximately 22% for Group 3 and 20% for Group 4. Inpatient within-day deteriorating ΔS-Cre or ΔS-Cre% above 0.2 mg/dL or 20%, respectively, is associated with all-cause mortality. Monitoring 24-hour S-Cre variation identifies acute kidney injury earlier than the conventional criteria.