Two decades of geological modelling have resulted in the ability to study single-well geological models at a sufficiently high resolution to generate synthetic well test responses from numerical simulations in realistic geological models covering a range of fluvial styles. These 3D subsurface models are useful in aiding our understanding and mapping of the geological variation (as quantified by porosity and permeability con-trasts) in the near-wellbore region. The building and analysis of these models enables many workflow steps, from matching well test data to improving history-matching. Well testing also has a key potential role in reservoir characterization for an improved understanding of the near-wellbore subsurface architecture in fluvial systems. Developing an understanding of well test responses from simple through increasingly more complex geological scenarios leads to a realistic, real-life challenge: a well test in a small fluvial reservoir. The geological well testing approach explained here, through a recent fluvial case study in South America, is considered to be useful in improving our understanding of reservoir performance. This approach should lead to more geologically and petrophysically consistent models, and to geologically assisted models that are both more correct and quicker to match to history, and thus, ultimately, to more useful reservoir models. It also allows the testing of a more complex geological model through the well test response.