The energy sources of geothermal systems in volcanic areas are cooling and solidifying intrusions in the crust. The nature of the heat transfer from the hot bodies to the groundwater is not known in detail. It is not known at what depth this interaction takes place, neither is the depth range of the water circulation. When geothermal modelers construct a model of a production field this issue is normally avoided. The heat sources are believed to be below the depth range of the model and are accounted for by choosing appropriate boundary conditions in the bottom layer of the model. Another standard practice in geothermal modeling is to drive the model using steady boundary conditions for a long period of time until it reaches a steady state, the so-called natural state, before production is simulated. This approach to model geothermal systems in volcanic areas has worked reasonably well for industrial purposes. However, a model that does not include the entire water circulation is not complete. Recently a well was drilled into magma in the Krafla Geothermal Field in North-Iceland at the depth of 2.1 km (FriA degrees leifsson et al., in: Proceedings World Geothermal Congress, 2010). This incident and an older example of a well in Nesjavellir in the Hengill Area, Southwest Iceland that was unexpectedly drilled into a very hot formation (C) in 1986 at the depth of 1800 m (Steingrimsson, in: Proceedings of Fifteenth Workshop on Geothermal Reservoir Engineering, 1990), have shown that the industry cannot overlook this imperfection in geothermal modeling. These examples from Krafla and Nesjavellir, along with other observations, indicate that the roots of the geothermal systems in the volcanic active zone in Iceland are at shallower depths than previously believed. Thus, the structure of existing numerical models has to be reconsidered. The heat sources have to be included into the models. Including the heat sources into the models also challenges the practice of driving the models to a steady state, since intrusions in the roots of volcanoes are not steady phenomena. Periods of high intrusion activity are known in volcanic systems in Iceland. In an improved modeling scheme, the heat sources should be included in any model, along with the entire water circulation, and the heat sources should be time dependent. We propose a simple conceptual model assuming that the heat sources of a geothermal systems are dykes at shallow depths ( km). Simple numerical models done in TOUGH2 show that dykes at shallow depths can indeed create a geothermal system similar to real systems.