This paper applies a 3-D nonuniform fast Fourier transform (NUFFT) migration method to image both free-space and buried targets from data collected by a ultra-wideband ground penetrating radar (GPR) system. The method incorporates the NUFFT algorithm into 3-D phase shift migration to evaluate the inverse Fourier transform more accurately and more efficiently than the conventional migration methods. Previously, the nonuniform nature of the wavenumber space required linear interpolation before the regular fast Fourier transform (FFT) could be applied. However, linear interpolation usually degrades the quality of reconstructed images. The NUFFT method mitigates such errors by using high-order spatial-varying kernels. The NUFFT migration method is utilized to reconstruct GPR images collected in laboratory. A plywood sheet in free space and a buried plexiglas chamber are successfully reconstructed. The results in 3-D visualization demonstrate the outstanding performance of the method to retrieve the geometry of the objects. Several buried landmines are also scanned and reconstructed using this method. Since the images resolve the features of the objects well, they can be utilized to assist the landmine discrimination.