Abstract The Vaca Muerta shale has been developed for oil and gas production since 2010 and to date nearly 500 wells have been drilled. The large amount of static and dynamic information from these wells has enabled fracture design and production strategy optimization. This paper details the methodology used to integrate all available data in 3D models, in order to understand the impact of rock properties in the production. The model was simulated using a commercial reservoir simulator, showing that hydraulic fractures are acting as a dual porosity system with a large conductivity (~10 D) connecting a low permeability matrix (~100 nD). We studied multiple wells in the history match (HM), using separator pressure and choke size as the control variables for the wells, and rates and pressures as comparison variables. A multi-segmented well approach was used to describe the pressure drop inside the well, and a vertical lift performance (VLP) table to describe the flow from the tubing all along to the separator including the wellhead choke. The static model included the seismic interpretation, stratigraphic framework, geomechanical and petrophysical characterization. Rock permeability, initial pore pressure and total fracture pore volume were calibrated with field measurements used as constraints in the HM process. Fracture conductivity degradation was introduced in the model to explain observed changes in the wells productivity. Laboratory tests are being designed to validate these hypotheses. We established early in the project that individual well HM were not unique. It was only through the HM of multiple wells that we were able to reduce the range of uncertainties affecting well performance (matrix permeability, initial water saturation and fracture height). This has given us a more reliable tool to obtain ultimate recovery estimation ranges. The described model showed a good prediction of a well with water lift problems, giving an accurate forecast for the incremental gas rate after a tubing diameter change. We concluded that the multi-segmented well model is a good representation of the water hold-up fraction behavior. This methodology enables us to integrate all the knowledge of the subsurface into a model that can be run in short simulation time (~30 minutes), allowing us to iterate quickly during the HM process. The model can be run for single wells or multiple wells and is flexible to adapt for new areas. We plan to use this methodology to design and monitor pilots in new blocks and to evaluate different development plans for existing projects.