Visual representations play an essential role in communicating molecular phenomena, particularly when it comes to education and outreach. Learners rely heavily upon visualizations to help them understand dynamic processes occurring over a wide range of temporal and spatial scales. This is reflected in the exponential increase of visualization resources available for both teaching and learning. This session will explore pitfalls and best practices in the depiction of molecular interactions by discussing strategies derived from practice-based heuristics and by highlighting research findings in this domain. We will examine the decision-making process involved in the design of molecular representations, in which considerations such as learning context and objectives, level of detail to include, and how to visually represent concepts where the evidence is lacking (or more hypothetical in nature), must all be balanced. We will also look at the role of artistic license in clarifying or abstracting complex phenomena when communicating to a non-expert audience. Through selective disclosure, the visualization designer can simplify a topic by stripping away all non-essential information to focus attention. The decision of "what to include" is a critical component of visualization in so far as the outcome has the potential to mislead (through oversimplification) or overwhelm the viewer (with extraneous detail). Whether molecular visualizations are driven by data (in the form of predictive or exploratory simulations) or informed by data (as in narratives or visual explanations), they are an important component of research and dissemination, which increasingly depends upon our capacity to innovate new forms of representation.