Understanding the Essential Requirements for Success in Structure-Based Drug Design We know that for any prediction the quality of the data used to build a model has a direct impact on the quality of the predictions from that model. For structure-based drug design (SBDD) the data is electron diffraction data collected on a protein-ligand crystal and the model is the protein-ligand structure or coordinates produced. While there can be some utility derived from most crystal structures, understanding the fine details of the binding interaction can only be done with the highest quality structures. Using the recently published Iridium data set, it will be shown that a high proportion of structures from the PDB could, in fact, entirely unsuitable for the task of understanding the details of binding for SBDD. Heuristics will be presented that will allow for a grading of the utility of protein-ligand crystal structures. The last part of this presentation will focus on a detailed evaluation of ligand conformations found in only the highest utility structures. We will present new methods that can dramatically improve ligand conformations, at little computational cost, and thus the utility of these structures for SBDD.