Background: Serum albumin concentration is a commonly available biomarker with prognostic value in many disease states. It is uncertain whether serum albumin concentrations are associated with incident end-stage renal disease (ESRD) independently of urine albumin-to-creatinine ratio (ACR). Methods: A longitudinal evaluation was performed of a population-based community-living cohort from the Reasons for Geographic and Racial Differences in Stroke (REGARDS) Study. Participants were ≥45 years of age at study entry and had serum albumin, creatinine, cystatin C and spot urine ACR measured at the baseline visit (n = 19 633). Estimated glomerular filtration rate (eGFR) was from the Chronic Kidney Disease Epidemiology Collaboration combined creatinine-cystatin C equation. Baseline serum albumin concentration was the predictor variable, and hazard ratios (HRs) for incident ESRD (from US Renal Data System linkage) were calculated in sequentially adjusted models. Results: Age at study entry was 63.9 ± 9.7 years, 62% of the participants were female and 40% were black. Mean eGFR at baseline was 83.3 ± 20.8 mL/min/1.73 m2. Over a median 8-year follow-up, 1.2% (n = 236) developed ESRD. In models adjusted for baseline eGFR, ACR and other ESRD risk factors, the HR for incident ESRD was 1.16 [95% confidence interval (CI) 1.01-1.33] for each standard deviation (0.33 g/dL) lower serum albumin concentration. The HR comparing the lowest (<4 g/dL) and highest quartiles (≥4.4 g/dL) of serum albumin was 1.61 (95% CI 0.98-2.63). Results were qualitatively similar among participants with eGFR <60 and ≥60 mL/min/1.73 m2, and those with and without diabetes. Conclusions: In community-dwelling US adults, lower serum albumin concentration is associated with higher risk of incident ESRD independently of baseline urine ACR, eGFR and other ESRD risk factors.