Ultrasonic guided wave imaging can potentially detect and localize damage over a large area with a sparse, or spatially distributed, array. However, conventional delay-and-sum imaging methods require knowledge of the propagation velocity and rely on direct-path propagation from the transmitting transducer, to a damage location, and finally to the receiving transducer, limiting the applicability of these techniques to relatively simple structures. A multipath guided wave imaging algorithm is presented here that leverages the large number of echoes and reverberations present in recorded ultrasonic waveforms to successfully detect and locate damage in geometrically complex structures. Multipath guided wave imaging not only allows imaging to be performed on structures with complex features (e.g. stiffeners, lap joints, and rivets) and inhomogeneous and anisotropic materials but also significantly improves image quality with far fewer sensors compared to conventional elliptical imaging. Experimental results from an aluminum plate and a carbon fiber-reinforced plastic composite panel are provided to illustrate imaging efficacy.