Warming is pushing the Earth system toward unfamiliar climate conditions, complicating predictions. Geological archives of past greenhouse climates provide essential tests for models under extreme forcing. We investigate how precipitation responded to extreme warmth during early Paleogene global warming events (66–47.8 million years ago) – a period considered a possible analogue for worst-case future scenarios. Here we compile global paleoclimate data and develop a multi-proxy approach that integrates sedimentary proxies - such as plant fossils, ancient soils and river deposits - providing constraints on global precipitation intermittency (seasonal and inter-annual variability) and intensity (rainfall rate). The data reveal wet or monsoonal polar regions, and aridity punctuated by intense rainfall at mid- and low-latitude continental interiors. This hydroclimate shift occurred 3 million years before and persisted 7 million years after the Paleocene-Eocene Thermal Maximum – the warmest period of the Cenozoic Era, suggesting that extreme warmth induces non-linearities in the hydrological cycle’s sensitivity to temperature increase. Polar humidity and mid-latitude aridity further indicate a departure from the expected wet-gets-wetter and dry-gets-drier response. Shifts towards aridity were decoupled from mean annual precipitation and driven by seasonal and interannual precipitation distribution, such as shorter wet season length and longer interannual rainfall recurrence interval. This highlights the importance of considering precipitation intermittency and intensity, as similar shifts may occur under future warming despite differences in boundary conditions.

This dataset was collected by compiling published literature on terrestrial temperature and precipitation during the Early Paleogene (66-47.8 Ma). Mean annual precipitation (MAP), mean annual temperature (MAT), and qualitative precipitation intermittency  and intensity values were used to assign a climate type with a distinct range of MAP and MAT values. Where quantitative information was not available, qualitative proxies were assigned a range of potential MAT and MAP values based on comparison to modern environments. For a more detailed explanation of the methods, please see the published article's supplementary information. 

Changes after 31st January 2024: Some new data were uploaded to the EECO table, and an additional column "geochronology" was added to all of the tabs.

Changes after 2nd February 2024: Corrected an incorrect lat/long.

Changes after 9th April 2024: The climate classification system was changed to be more similar to the Koppen-Geiger system. The new system is described in the first tab of the data. Furthermore, the reference for the Tornillo Basin in each tab was updated based on a publication that was previously unknown to us.

Changes after 27th May 2024: Quantitative uncertainty values were added for MAT and MAP. Additionally, a table was added to give a range of values for qualitative descriptors of MAT and MAP. 

Changes after 30th September 2024: Updated the document name to correspond to the most recent manuscript revision. 

Changes after 22nd April 2025: The DeepMIP_Combined and ProxyModelComparison spreadsheets were added because the manuscript has been updated to include a model-data comparison. Furthermore, additional citations were added based on new data that has been released since the last version. 

Changes after 25th August 2025: The title was updated to correspond to the newest version of the manuscript. The PaleogeneClimate.xlsx file was replaced to include the properly named "precipitation intensity" column, with the README correspondingly updated as well. 

Changes after Aug 26, 2025: There was a typo in the Paleogene climate file - "precipitation intermittency" was duplicated when one of the columns needs to read "precipitation intensity."

Changes after 27th October 2025: Supplemental data was combined into a single file at the request of the journal. The README was correspondingly updated as well.