Ground-motion prediction equations (GMPEs) play a crucial role for estimating the seismic hazard in any region using either a deterministic or a probabilistic approach. Indeed, they represent a reliable and fast tool to predict strong ground motion, given source and propagation parameters. In this article, we estimated GMPEs for the South Korea peninsula. GMPEs were computed for peak ground displacement, peak ground velocity, peak ground acceleration, and spectral accelerations (damping at 5%) at 13 different periods from 0.055 to 5 s. We analyzed data from 222 earthquakes recorded at 132 three-component stations of the South Korea Seismic Network, from 2007 to 2012, with local magnitude ranging between 2.0 and 4.9 and epicentral distances varying from 1.4 to ~600 km. A nonlinear mixed effects technique is used to infer the GMPE coefficients. This technique includes both fixed and random effects and accounts for both inter- and intraevent dependencies in the data. Station-specific corrective coefficients were estimated by a statistical approach and were included in the final ground-motion prediction model. Finally, predictions for peak ground acceleration and spectral acceleration are compared with observations recorded for an M L 5.1 earthquake that occurred in 2014, the data for which were not included in the modeling. Online Material: Figures showing final ground-motion prediction equation models versus observations, and intra- and interevent residuals.