We present a suite of strain rate models for the western United States based on geologic and geodetic data. The geologic data consist of Quaternary fault‐slip rates and the geodetic data consists of a new compilation of Global Positioning System (GPS) velocities derived from continuous, semicontinuous, and campaign measurements. We remove postseismic deformation from the GPS time series in order for our geodetic strain rate model to best capture the interseismic strain accumulation rate. We present models based on either geologic or geodetic data, but also create a hybrid model. Although there are some differences between the models, the large‐scale features are the same, with the noticeable exception for the Pacific Northwest where interseismic strain is naturally more distributed than the long‐term strain release. We also present a map of earthquake rate densities based on mainshocks, and the result has similar spatial features similar to the strain rate models (at least in the southwestern United States). We perform a general correlation analysis between strain rate and seismicity rate (south of Cascadia) and find a change in linearity between seismicity and strain rates from slow to faster deforming areas with seismicity rates relatively lower for the latter. The extent of that change depends a bit on assumptions made on the declustering and completeness of the catalog, but the finding of a change in slope is robust across the different strain rate models. Linearity for all areas is only expected when Gutenberg–Richter parameters and parameters involved in the conversion from strain to moment rate are uniform across the study area. We discuss these qualifications, but find no single satisfactory explanation for our observation. Moreover, when considering a rather short time and space, theoretical considerations of sampling from a power‐law distribution actually predict there to be a power law instead of a linear relationship, generally consistent with our observation.