A new empirical model for predicting ground-level geomagnetic perturbations has been developed. This model is based on global measurements of the magnetic field at multiple stations in the Northern Hemisphere collected over an 8 year period, along with the simultaneous measurements of the interplanetary magnetic field (IMF). Variations in ionospheric conductivity are implicitly contained in the measurements used in the model's construction, including the solar F-10.7 index. Provided with the IMF, solar wind velocity, dipole tilt angle (for season), and F-10.7 index, this model computes all three vector components of the magnetic perturbations at specified locations. The model results are consistent with the corresponding maps of the ionospheric electric potential. Interestingly, maps of the vertical component have patterns that resemble maps of the overhead, ionospheric field-aligned currents. Comparisons of model calculations with measurements at different locations show very good results, particularly at low frequencies. There are random variations at higher frequencies that are not reproduced well with the model, but they tend to occur in proportion to the predicted levels. This model could be useful for providing regional forecasts of geomagnetic activity with an approximately 1 h lead time. Citation: Weimer, D. R. (2013), An empirical model of ground-level geomagnetic perturbations, Space Weather, 11, 107-120, doi:10.1002/swe.20030