The idea of NMR crystallography was conceived as soon as NMR spectroscopy was invented. Over the years, several efforts have been devoted to the development of NMR tools to complement X-ray diffraction results. Many of the NMR-related observables are short-ranged in nature, but the paramagnetic ones are not. Therefore, paramagnetism-based restraints are in principle suitable to assess spatial relationships among molecules. We will here review some of the underlying concepts and see how they apply to the problem of obtaining structural information on molecules in the lattice, and on the lattice itself. We will provide positive examples and discuss the negative issues that still haunt paramagnetism-based NMR crystallography.