A coupled WRF–ROMS–SWAN framework is used to investigate the long-term wave characteristics of the Bohai Sea through a 20-year hindcast from 2002 to 2022. The model results are validated against CMEMS current data, field wave measurements, and CCMP wind products, showing good agreement with the reference datasets. Three spatial interpolation methods are compared for reconstructing the wave time series at discrete locations, and distance-squared weighted interpolation (DSWI) provides the best overall performance for the present dataset. The reconstructed records show clear seasonal and interannual variations in the significant wave height, average wavelength, and spectral peak period. Spatially, larger mean and extreme wave parameters occur mainly in the central and eastern offshore areas, while the enclosed inner bays are generally calmer. Weibull, Gumbel, Pearson Type III, and Exponential distributions are further applied to estimate wave conditions for return periods of 1–100 years. The Exponential distribution gives the highest long-period estimates, whereas the Weibull distribution gives the lowest. Wave roses also show clear differences in prevailing wave directions among stations. The results provide long-term wave information for offshore engineering and marine hazard assessment in the Bohai Sea.