Background Reduced physical function, a representative phenotype of aging, has been associated with cardiovascular disease (CVD). However, few studies have comprehensively investigated its association with composite and individual CVD outcomes in community‐dwelling older adults and its predictive value for CVD beyond traditional risk factors. Methods and Results We studied 5570 participants (mean age 75 [SD 5] years, female 58%, Black 22%) at visit 5 (2011–2013) of the ARIC (Atherosclerosis Risk in Communities) study. Physical function was evaluated with the Short Physical Performance Battery (SPPB), which incorporates a walk test, chair stands, and balance tests. The SPPB score was modeled categorically (low [0–6], intermediate [7–9], and high [10–12]) and continuously. We assessed the associations of SPPB score with subsequent composite (coronary heart disease, stroke, or heart failure) and individual CVD outcomes (components within composite outcome) using multivariable Cox models adjusting for major CVD risk factors and history of CVD. We also evaluated improvement in C‐statistics by adding SPPB to traditional CVD risk factors in the Pooled Cohort Equation. Among the study participants, 13% had low, 30% intermediate, and 57% high SPPB scores. During a median follow‐up of 7.0 (interquartile interval 5.3–7.8) years, there were 930 composite CVD events (386 coronary heart disease, 251 stroke, and 529 heart failure cases). The hazard ratios of composite CVD in low and intermediate versus high SPPB score were 1.47 (95% CI, 1.20–1.79) and 1.25 (95% CI, 1.07–1.46), respectively, after adjusting for potential confounders. Continuous SPPB score demonstrated independent associations with each CVD outcome. The associations were largely consistent across subgroups (including participants with prevalent CVD at baseline). The addition of SPPB to traditional CVD risk factors significantly improved the C‐statistics of CVD outcomes (eg, ΔC‐statistic 0.019 [95% CI, 0.011–0.027] for composite CVD). Conclusions Reduced physical function was independently associated with the risk of composite and individual CVD outcomes and improved their risk prediction beyond traditional risk factors in community‐dwelling older adults. Although confirmatory studies are needed, our results suggest the potential usefulness of SPPB for classifying CVD risk in older adults.