We propose Direct Position Estimation (DPE) utilizing non-line-of-sight (NLOS) GPS signals for urban navigation. In urban environments, buildings reflect GPS signals, leading to reception of NLOS GPS signals. In addition, buildings obstruct GPS signals, resulting in reduced GPS signal availability. We treat NLOS GPS signals as additional line-of-sight (LOS) GPS signals to virtual satellites at mirror-image positions. We calculate these satellite mirror-image positions and velocities using knowledge of building reflection surfaces estimated from available three-dimensional (3D) maps. We then create expected signal receptions to include NLOS GPS signal information at multiple potential navigation candidates. Following that, we use Direct Position Estimation (DPE) to obtain the Maximum Likelihood Estimate (MLE) navigation solution. The MLE navigation solution is the navigation candidate with the expected signal reception that produced the highest overall correlation against the actual received signal. We implemented DPE utilizing NLOS GPS signals on our software platform - PyGNSS. We conducted experiments in front of the 53 m by 40 m wind tunnel located at NASA Ames Research Center in Mountain View, California. The surface of the wind tunnel’s air intake valve is a dense metal mesh of grid size 1.8 cm, a reflector of GPS signals. We demonstrated through our experiment, an improvement in horizontal positioning accuracy by 40 m in comparison to conventional GPS positioning using scalar tracking.