The Geysers geothermal field in Northern California, which has been actively exploited since the 1960s, is the world’s largest geothermal field. The continuous injection of fluids and the consequent stress perturbations induce seismicity that is clearly felt in the surrounding communities. In order to evaluate seismic hazard due to induced seismicity and the effects of seismicity rate level on the population and buildings in the area, reliable ground‐motion prediction equations (GMPEs) must be developed. This paper introduces the first GMPEs specific for The Geysers area in terms of peak ground velocity (PGV), peak ground acceleration (PGA), and 5% damped spectral acceleration SA(T) at T=0.2 s, 0.5 s, and 1.0 s. The adopted non‐linear mixed‐effect regression technique to derive the GMPE includes both fixed and random effects, and it permits to account for both inter‐event and intra‐event dependencies in the data. Site‐specific effects are also estimated from the data and are corrected in the final ground‐motion model. We used data from earthquakes recorded at 29 stations of the Berkeley‐Geysers network during the period September 2007 through November 2010. The magnitude range is 1.3≤Mw≤3.3, whereas the hypocentral distances range between 0.5 km and 20 km. The comparison of our new GMPE for The Geysers with a standard model derived in a different tectonic context shows that our model is more robust when predictions have to be made for induced earthquakes in this geothermal area.