The Okataina Volcanic Centre (OVC) in the Taupo Volcanic Zone (TVZ) is New Zealand's most recently active caldera and lava dome complex, erupting 85 km3 of magma equivalent in the past 21 ka. Successive caldera collapses over the past ∼550 kyr have created a complex crustal structure through which fluids now migrate, evidenced by hot springs and several geothermal fields located predominantly around the caldera margins. While recent magnetotelluric (MT) modelling suggests that high-temperature upwelling fluids originate from mid-crustal locations at the margins of an inferred magmatic complex, it is still uncertain to what degree various factors influence these fluid pathways from source to surface. Here, we use heat and fluid flow models to explore the roles that simplified geological structures, topography, and deep, localised heat sources play in determining the locations of upflow at the OVC.To isolate and understand individual processes that modulate fluid flow, we created TOUGH2 heat and fluid flow models that encompass a 35 by 39 km2 area extending beyond the topographic boundaries of the OVC calderas. An average heat input of 700 mW/m2 was set at the base of the models at 5 km depth, either as a uniform hot plate or as discrete zones of higher heat input with locations derived from MT models. The upper boundary of the models followed water table elevation, which is a slightly muted reflection of topography. We constructed a suite of models, ranging from uniform rock properties to more complex permeability variations that correspond to basement/volcanic rocks and large-scale faults inferred from gravity data and surface mapping. Models ran for 50 kyr, the approximate time since the most recent caldera-forming event. Elevated model temperatures were compared with low-resistivity zones from shallow (<500 m) Direct Current (DC) apparent resistivity values and deeper sensing MT resistivity models.Our models suggest that topography and localised basal heat sources are the largest influences on the locations of geothermal upflows. In the western half of the study area, observed upflows beneath Waimangu and Tikitere can be modelled with localised heat sources and topography alone. Upflow under Rotoma to the north-east is not replicated by any of our suite of models, suggesting that unmapped heat sources are important in this area. In the east of the study area, modelled upflow is dominated by present-day topography which is strongly influenced by <25ky Tarawera and Haroharo Volcanic Complexes and < 1ky Tarawera River valley. Our models suggest that strong permeability contrasts associated with large-scale faults would have effects on regional-scale fluid circulation that are not observed in real life, and that the coincidence of OVC topographic boundaries and geothermal systems is due to the boundaries' also being topographic lows rather than fluid conduits.