Despite continuing advances in biophysical techniques it is still not possible to directly observe the energetic and dynamic properties of individual atoms in biomolecular systems using experimental approaches. As a consequence, everything we know (or think we know) regarding biomolecular systems at an atomic level is to a greater or lesser degree a model. The question then becomes to what extent can the models used to interpret experimental data be trusted? For example, while the overall structure of a protein may be resolved in near atomic detail, the position, orientation and/or conformation of a small molecular ligand (cofactor, substrate, inhibitor etc.) bound to such a protein is often much less certain.1 Electrostatics are generally ignored in structure refinement. In other cases there is systematic bias in how data is used or interpreted. The variation in the area per lipid in theoretical calculations of membrane systems is much less than the variation in the experimental data on which they are based.2 This is not only a problem for experimentalists but represent a fundamental challenge to theoreticians attempting to validate computational models.3 4 Using a range of examples the talk will illustrate how the theoretical or structural model models we use can bias our interpretation of experimental data, but how when used appropriately molecular simulations, provide a unique means to test and challenge our interpretation of experimental data leading to increased certainty in some cases, new insights into the true nature of biomolecular systems in action in others.5 6