Background: Dapagliflozin reduced the risk of kidney failure in patients with chronic kidney disease with and without type 2 diabetes in the DAPA-CKD trial. In this pre-specified analysis, we assessed the effect of dapagliflozin on the rate of change in estimated glomerular filtration rate (eGFR)—ie, the eGFR slope. Methods: DAPA-CKD was a randomised controlled trial that enrolled participants aged 18 years or older, with or without type 2 diabetes, with a urinary albumin-to-creatinine ratio (UACR) of 200–5000 mg/g, and an eGFR of 25–75 mL/min per 1·73m². Participants were randomly assigned (1:1) to oral dapagliflozin 10 mg once daily or placebo, added to standard care. In this pre-specified analysis, we analysed eGFR slope using mixed-effect models with different slopes from baseline to week 2 (acute eGFR decline), week 2 to end of treatment (chronic eGFR slope), and baseline to end of treatment (total eGFR slope). DAPA-CKD is registered with ClinicalTrials.gov, NCT03036150, and is now complete. Findings: Between Feb 2, 2017, and April 3, 2020, 4304 participants were recruited, of whom 2152 (50%) were assigned to dapagliflozin and 2152 (50%) were assigned to placebo. At baseline, the mean age was 62 years (SD 12), 1425 (33·1%) participants were women, 2906 (67·5%) participants had type 2 diabetes. The median on-treatment follow-up was 2·3 years (IQR 1·8–2·6). From baseline to the end of treatment, dapagliflozin compared with placebo slowed eGFR decline by 0·95 mL/min per 1·73 m² per year (95% CI 0·63 to 1·27) in the overall cohort. Between baseline and week 2, dapagliflozin compared with placebo resulted in an acute eGFR decline of 2·61 mL/min per 1·73 m² (2·16 to 3·06) in patients with type 2 diabetes and 2·01 mL/min per 1·73 m² (1·36 to 2·66) in those without type 2 diabetes. Between week 2 and end of treatment, dapagliflozin compared with placebo reduced the mean rate of eGFR decline by a greater amount in patients with type 2 diabetes (mean difference in chronic eGFR slope 2·26 mL/min per 1·73 m² per year [1·88 to 2·64]) than in those without type 2 diabetes (1·29 mL/min per 1·73 m² per year [0·73 to 1·85]; p_interaction=0·0049). Between baseline and end of treatment, the effect of dapagliflozin compared with placebo on the decline of total eGFR slope in patients with type 2 diabetes was 1·18 mL/min per 1·73 m² per year (0·79 to 1·56) and without type 2 diabetes was 0·46 mL/min per 1·73 m² per year (–0·10 to 1·03; p_interaction=0·040). The total eGFR slope was steeper in patients with higher baseline HbA_1c and UACR; the effect of dapagliflozin on eGFR slope was also more pronounced in patients with higher baseline HbA_1c and UACR. Interpretation: Dapagliflozin significantly slowed long-term eGFR decline in patients with chronic kidney disease compared with placebo. The mean difference in eGFR slope between patients treated with dapagliflozin versus placebo was greater in patients with type 2 diabetes, higher HbA_1c, and higher UACR. Funding: AstraZeneca.