To investigate the association between serum albumin levels and cause-specific mortality among community-dwelling older adults. This cohort study was based on data obtained from the government-sponsored Annual Geriatric Health Examination Program for the older adults in Taipei City between 2006 and 2010. The study sample consisted of 77,531 community-dwelling Taipei citizens (≥65 years old). Mortality was determined by matching the participants' medical records with national death files. Serum albumin levels were categorized into <3.6, 3.6–3.7, 3.8–3.9, 4.0–4.1, 4.2–4.3, and ≥4.4 g/dL. Cox proportional hazards regression models were used to evaluate the association between albumin levels and cause-specific mortality. Spline regression was used to calculate the risk of mortality associated with albumin levels, modeled as continuous variables. Community-dwelling older adults had a mean albumin level of 4.3 g/dL, which significantly reduced by age. Compared to albumin levels ≥4.4 g/dL, mildly low albumin levels (4.2–4.3 g/dL) were associated with an increased mortality risk (hazard ratio [HR]: 1.16, 95% confidence interval [CI]: 1.05–1.28 for all-cause mortality), and albumin levels <4.2 g/dL were associated with significantly higher rates of all-cause, cancer, cardiovascular, and respiratory mortalities. In the spline regression, the curve of mortality risk was relatively flat at an albumin level ≥4.4 g/dL, and the mortality risk gradually increased as the albumin level declined. Albumin levels ≥4.4 g/dL were associated with better survival among community-dwelling older adults, and mortality risk increased as the albumin level decreased.