This article addresses, by means of computation and advanced experiments, one of the key challenges of NMR crystallography, namely the assignment of individual resonances to specific sites in a crystal structure. Moreover, it shows how NMR can be used for crystal structure validation. The case examined is form B of terbutaline sulfate. CPMAS 13 C and fast MAS 1 H spectra have been recorded and the peaks assigned as far as possible. Comparison of 13 C chemical shifts computed using the CASTEP program (incorporating the Gauge Including Projector Augmented Wave principle) with those obtained experimentally enable the accuracy of the two distinct single-crystal evaluations of the structure to be compared and an error in one of these is located. The computations have substantially aided in the assignments of both 13 C and 1 H resonances, as has a series of two-dimensional (2D) spectra (HETCOR, DQ-CRAMPS and proton-proton spin diffusion). The 2D spectra have enabled many of the proton chemical shifts to be pinpointed. The relationships of the NMR shifts to the specific nuclear sites in the crystal structure have therefore been established for most 13 C peaks and for some 1 H signals. Emphasis is placed on the effects of hydrogen bonding on the proton chemical shifts.