Historical seismic intensity data are useful for myriad reasons, including assessment of the performance of probabilistic seismic hazard assessment (PSHA) models and corresponding hazard maps by comparing their predictions to a dataset of historically observed intensities in the region. To assess PSHA models for California, a long and consistently interpreted intensity record is needed. For this purpose, the California Historical Intensity Mapping Project (CHIMP) has compiled a dataset that combines and reinterprets intensity information that has been stored in disparate and sometimes hard‐to‐access locations. The CHIMP dataset also includes new observations of intensity from archival research and oral history collection. Version 1 of the dataset includes 46,502 intensity observations for 62 earthquakes with estimated magnitudes ranging from 4.7 to 7.9. The 162 yr of shaking data show observed shaking lower than expected from seismic hazard models. This discrepancy is reduced, but persists, if historical intensity data for the largest earthquakes are smoothed to reduce the effects of spatial undersampling. Possible reasons for this discrepancy include other limitations of the CHIMP dataset, the hazard models, and the possibility that California seismicity throughout the historical period has been lower than the long‐term average. Some of these issues may also explain similar discrepancies observed for Italy and Japan.