Strainmeters can record offsets coincident with earthquakes, but how much these represent strain changes from elastic rebound, and how much they are contaminated by local effects, remains an open question. To study this, we use a probabilistic detection method to estimate coseismic offsets on nine borehole strainmeters (BSMs) operated by the Plate Boundary Observatory (PBO) in southern California, from 34 earthquakes with a wide range of magnitudes and distances. In general, the offsets estimated for the BSM data differ substantially from the static strain predicted by elastic dislocation theory, which is well supported by other techniques, though 10% of the observed offsets agree well with theory. For one earthquake, the BSM offsets significantly disagree with collocated long‐base laser strainmeter data. Comparisons with collocated seismic data provide strong evidence that the absolute errors between the observed and predicted strains scale with the level of seismic energy density but also that relative errors (normalized by the model size) do not. We conclude that apparent strain offsets are induced by seismic waves, occurring presumably because of irreversible deformation, whether in the rock or cementing materials close to the BSMs, or in the instruments themselves. Coseismic offsets seen in PBOBSM data should therefore be viewed with caution before being used as a measure of large‐scale coseismic deformation.
Online Material: Tables of offset estimates and predictions, error estimates, strike sensitivities, and figures of observed versus predicted offsets.