The inverse Gaussian distribution is a probability distribution used for modeling continuous, positive data. It is particularly useful for analyzing survival times and representing the distribution of particulate matter 2.5 (PM2.5) in environmental studies. For PM2.5, calculating percentiles can help establish thresholds or benchmarks for different air quality levels. For instance, the 95th percentile might represent high pollution events, indicating a PM2.5 level that exceeds 95% of the observed data. Therefore, the aim of this study is to construct confidence intervals for the percentiles of the inverse Gaussian distribution using four methods: the generalized confidence interval, the adjusted generalized confidence interval, the Bayesian method, and the highest posterior density confidence interval. Monte Carlo simulations were performed to evaluate the performance of these intervals, assessing them in terms of coverage probability and average length. The results indicate that the generalized confidence interval performs well with small sample sizes, whereas the adjusted generalized confidence interval method is more suitable for large sample sizes. Furthermore, these methods were applied to PM2.5 data from Bangkok, Thailand, to demonstrate their consistency with both real-world and simulated data.