Traditional multimodel estimates of precipitation intensity-duration-frequency (IDF) curves based upon the mean or median of models’ IDF estimates suffer from large estimation uncertainty. Such large uncertainties around multimodel IDF estimates impair their utility in planning and adaptation efforts. In this study, assuming that each climate model is one representation of the underlying data generating process, i.e., the Earth system, we propose a novel extension of current methods through pooling of model data, which has 3 steps: (i) evaluate the spatial and temporal variability of the annual maximum precipitation (AMP) in the historical simulations of climate models against that in the observations (ii) bias-correct and pool the historical and future AMP data of reasonably performing models, and (iii) compute IDF estimates in a nonstationary framework from pooled model data. Pooling enhances fitting of the extreme value distribution to the data and assumes that data from reasonably performing models represent samples from the “true” underlying data-generating distribution. Using Monte Carlo experiments performed on synthetic data, we show that return periods derived from pooled data have smaller biases (difference between the observed and the median of the estimated return periods) and lesser uncertainty (interquartile range of the estimated return periods). When applied to 24-hr AMP in the North American Coordinated Regional Climate Downscaling Experiment (NA-CORDEX) models over the Susquehanna watershed and Florida peninsula, our method identifies statistically significant future changes in 24-hr precipitation intensity-frequency (PIF) estimates at more stations compared to median-based PIF estimates for all return periods examined. Our analysis suggests that almost all stations over the Susquehanna and at least two-thirds of the stations over the Florida peninsula will observe significant increases in 24-hr precipitation for 2–100-yr return periods.