BACKGROUND: Stroke is a common fatal and disabling disease in the elderly. We investigated the correlation and potential benefits of dietary intake of flavonoid compounds and their subclasses in elderly stroke patients. METHODS: This study analyzed data from the National Health and Nutrition Examination Survey (NHANES) for the cycles 2007-2010 and 2017-2018. Dietary flavonoids intake was calculated based on food frequency questionnaires (FFQ) through a 24-hour dietary recall. To explore the relationship between flavonoids intake and their subclasses with stroke prevalence, we employed multivariable logistic regression models, adjusting for relevant covariates. Restricted cubic splines (RCS) and generalized additive models (GAMs) were applied to explore potential nonlinear relationships between flavonols subclasses and stroke prevalence. Multivariable logistic regression models were conducted both for the overall population and stratified by flavonols subclasses intake. Additionally, we utilized Bayesian kernel machine regression (BKMR) to evaluate the overall effect of flavonols intake levels on stroke status, accounting for potential interaction and correlation between subclasses. Given the high correlation among flavonols subclasses, we incorporated a hierarchical variable selection approach using a Markov Chain Monte Carlo (MCMC) algorithm. Finally, we analyzed the XGBoost model by applying SHAP (SHapley Additive exPlanations) values to assess the specific contribution of different flavonoids intake to stroke. RESULTS: Including 3,806 elderly stroke patients, the study revealed an inverse relationship between dietary flavonoids and their subclasses and stroke prevalence. After adjusting for potential confounders, it was found that higher quartiles of flavonols intake were associated with lower stroke prevalence. Specifically, with each unit increase in flavonols (Q4) intake, the odds of stroke in the elderly decreased by 61 % (OR=0.390, 95 % CI [0.209-0.728]; P = 0.005). Similar results were observed for the subclasses of flavonols. Subgroup analyses indicated that age and poverty index ratio (PIR) were effect modifiers in the relationship between flavonols intake and stroke. We further examined the intake levels of dietary flavonols subclasses such as isorhamnetin, kaempferol, myricetin, and quercetin and their association with stroke status stratified by population characteristics. In addition to age and PIR, hyperlipidemia and body mass index (BMI) were found to be the most common significant influencing factors in the relationship between flavonols subclasses and stroke prevalence. Furthermore, RCS revealed a "U"-shaped nonlinear relationship between flavonols, including their quercetin and kaempferol subclasses, and stroke, whereas the relationship between stroke and myricetin was linear. Our study also assessed the overall impact of dietary flavonols subclasses on stroke in the elderly and the interrelationships among these subclasses. The results consistently indicated a negative joint effect of flavonols subclass mixtures on the stroke status in the elderly. When evaluating the impact of individual flavonols subclasses on stroke status, a potential dose-response relationship was observed, with increasing intake of myricetin being associated with a decreased risk of stroke. Furthermore, SHAP plots show that flavonols intake is predominantly associated with negative SHAP values, indicating that higher dietary flavonols intake corresponds to a lower predicted stroke risk. CONCLUSION: These results emphasize that adhering to an increased dietary intake of flavonoid compounds, particularly flavonols and their subclasses such as myricetin, is inversely associated with the prevalence of stroke among the U.S. elderly population. This offers potential benefits for stroke patients, especially among elderly individuals aged 60-70 and those with higher incomes.