Objective: Prospective studies, mostly from Europe and North America, suggest that serum 25-hydroxyvitamin D (25(OH)D) is inversely associated with mortality and cardiovascular disease (CVD) risk. Data from other regions are limited, and threshold levels for adverse cardiovascular outcomes are uncertain. We examined serum 25(OH)D as a predictor of total mortality and cardiovascular outcomes in an Australian cohort. Design: A 20-year, community-based cohort study. Patients: Participants in the 1994/1995 Busselton Health Survey (n = 3946, baseline age 25-84 years). Measurements: Baseline serum 25(OH)D and mortality and cardiovascular outcomes to 2014 obtained by record linkage. Results: The mean serum 25(OH)D concentration was 60.6 ± 18.0 nmol/L. During 20-year follow-up (excluding the first 2 years), 889 participants died (including 363 from CVD) and 944 experienced a CVD event (including 242 with heart failure). In the full cohort, controlling for Framingham risk score variables, higher baseline 25(OH)D was associated with significantly reduced all-cause mortality (adjusted HR 0.83 per SD increment of 25(OH)D, 95% CI 0.77-0.90), CVD death (HR 0.85, 95% CI 0.74-0.96) and heart failure (HR 0.81, 95% CI 0.69-0.94), but not CVD events (HR 0.99, 0.92-1.07). In restricted cubic spline regression models, serum 25(OH)D below 65 and 55 nmol/L was associated with higher total mortality and higher CVD mortality/heart failure, respectively. In participants without CVD at baseline (n = 3220), results were similar, but hazard ratios were attenuated and associations with CVD mortality no longer significant. Conclusions: In an Australian community-based cohort, baseline vitamin D levels below 55-65 nmol/L are predictive of all-cause mortality, CVD death and heart failure.