One of the prime merits of NMR as a tool for lead finding in drug discovery research is its sensitivity and robustness to detect weak protein-ligand interactions. This sensitivity allows to build up ligands for a given target in a modular way, by a fragmentbased approach. In this approach, two ligands are seperately identified which bind to the target protein generally weakly, but at adjacent binding sites. In a next step, they are chemically linked to produce a high-affinity ligand. This review discusses methods to detect “second-site” ligands that bind to a protein in the presence of a “first-site” ligand, and methods to elucidate structural details on the spatial orientation of both ligands, so that chemical linkage is based on a large piece of experimental information. Published examples from second-site screening and linker design are summarized, and are complemented by previously unpublished in-house examples.