<p>Accurate estimation of wind energy potential is crucial for the transition to wind power. For numerous years, extensive research has been conducted to assess wind energy potential. However, determining the optimal and most reliable distribution for wind speed modeling remains challenging. Most wind resource assessments rely on the 2p-Weibull distribution; yet, it is inappropriate to use the same distribution for every location because of the varying nature of wind speed geographically and temporally. In this study, ten statistical distributions: Normal, Log-normal, 3p-Weibull, 2p-Weibull, Gamma, Rayleigh, Burr, Log-logistic, Inverse-Gaussian, and Inverse-Gamma are employed to fit the wind speed data recorded from ten sites across Pacific Island Countries (PICs). Model performance was evaluated using ten model evaluation criteria: R², COE, RMSE, MAE, MAPE, KS, CVM, AD, AIC, and BIC. Additionally, four estimation methods were employed to assess statistical distribution and model evaluation criteria: MLE, MME, QME, and MGE. Furthermore, the best estimation method is identified using the TOPSIS method. The performance of distributions was ranked on a scale of 1(best) to 3(third best); the findings reveal that the 3p-Weibull provides the best result for 42.5%, and MGE performed best for most of the studied sites, whereas the traditional 2p-Weibull, Burr, or Rayleigh were better candidates for some sites. Overall, 3p-Weibull emerges as the best-fitted distribution across the studied sites, though its performance is not universal. These findings emphasize the importance of considering multiple statistical distributions and estimation methods for accurate wind speed modeling. Additionally, wind power densities are estimated for each site using the best-fitted distribution, which showed discrepancies between the estimated and actual wind power densities, highlighting the importance of an appropriate distribution and estimation method.</p>