<p>The study of extreme precipitation events has become a major research topic due to its importance in a climate change context. The determination of extreme events and their study usually depend on the estimation of daily percentiles. Therefore, this research evaluates the performance of different approaches and methods to estimate daily precipitation percentiles. To achieve this, simulations of five different climate regimes were conducted to evaluate each method's performance. Four distinctive factors were considered: the percentile to be estimated, the usage of a wet day or all days, the estimation method, and the usage of a smoothing technique after the estimation. Regarding the usage of wet days, we found that a wet-day threshold of 0&#xa0;mm generally performed better than a 1&#xa0;mm threshold. Moving window approaches yielded better results than methods using only the calendar day, leveraging larger data subsets. Smoothing techniques, particularly Generalized Additive Models (GAM), significantly improved performance. The choice of wet-day definition and percentile depends on research goals, affecting threshold levels and the number of extreme events, which influence statistical analyses. Higher percentiles showed decreased method performance, being less representative. Given potential biases with wet-day thresholds and minimal performance differences, we recommend using all days for percentile estimation in future research. For this case, the moving empirical percentile estimation method with GAM smoothing is advised. Nevertheless, optimal techniques may vary by climate. Lastly, the differences in the annual frequency of extreme events index derived from ERA5-Land data using different percentile estimation methods were analyzed. The findings suggest that the choice of percentile estimation method has a greater impact on analyzing time series variability but less influence on linear trend analysis.</p>