We demonstrate the use of POT (Peaks over Threshold) models to evaluate distribution of extreme winds within near-coastal regions of the United States. Initially, we use maximum likelihood methods to estimate the return levels for various long-range return periods for each coastal region. Next, we show that the return levels depend on climate variables such as ENSO and NAO. Finally, we demonstrate the usefulness of the Bayesian approach to POT modeling using WinBUGS software to model the relationship between climate variables and hurricane intensity. Additionally, we show that the Bayesian approach is useful for extending the historical record, managing measurement error, imputing missing values, and simulating future hurricane intensities.