This review will focus on the construction, refinement, and validation of G Protein-coupled receptor models for the purpose of structure-based virtual screening. Practical tips and tricks derived from concrete modeling and virtual screening exercises to overcome the problems and pitfalls associated with the different steps of the receptor modeling workflow will be presented. These examples will not only include rhodopsin-like (class A), but also secretine-like (class B), and glutamate-like (class C) receptors. In addition, the review will present a careful comparative analysis of current crystal structures and their implication on homology modeling. The following themes will be discussed: i) the use of experimental anchors in guiding the modeling procedure; ii) amino acid sequence alignments; iii) ligand binding mode accommodation and binding cavity expansion; iv) proline-induced kinks in transmembrane helices; v) binding mode prediction and virtual screening by receptor-ligand interaction fingerprint scoring; vi) extracellular loop modeling; vii) virtual filtering schemes. Finally, an overview of several successful structure-based screening shows that receptor models, despite structural inaccuracies, can be efficiently used to find novel ligands.