The (15)N chemical shift tensor is shown to be extremely sensitive to lattice structure and a powerful metric for monitoring density functional theory refinements of crystal structures. These refinements include lattice effects and are applied here to five crystal structures. All structures improve based on a better agreement between experimental and calculated (15)N tensors, with an average improvement of 47.0 ppm. Structural improvement is further indicated by a decrease in forces on the atoms by 2-3 orders of magnitude and a greater similarity in atom positions to neutron diffraction structures. These refinements change bond lengths by more than the diffraction errors including adjustments to X-Y and X-H bonds (X, Y = C, N, and O) of 0.028 ± 0.002 Å and 0.144 ± 0.036 Å, respectively. The acquisition of (15)N tensors at natural abundance is challenging and this limitation is overcome by improved (1)H decoupling in the FIREMAT method. This decoupling dramatically narrows linewidths, improves signal-to-noise by up to 317%, and significantly improves the accuracy of measured tensors. A total of 39 tensors are measured with shifts distributed over a range of more than 400 ppm. Overall, experimental (15)N tensors are at least 5 times more sensitive to crystal structure than (13)C tensors due to nitrogen's greater polarizability and larger range of chemical shifts.