Installation of Ground-Source Heat Pump (GSHP) system is gradually increasing in Japan. Thermal conductivity of the formation in Japan is smaller than that in continental area consisting of rock which is older than Quaternary because major cities and towns in Japan are located on Quaternary plains and basins. However, groundwater of Quaternary is actively flowing, and the heat exchange rate depends on the existence of groundwater and its flow rate. Therefore, GSHP suitability mapping has been carried out based on hydrological parameters such as groundwater flow velocity, groundwater temperature and hydraulic head in Japan. In this study, formularization of the relationship between heat exchange rate and its influential factor was conducted and the method to predict heat exchange rate from the results of groundwater flow and heat transport simulation using response surface methodology. First, a 3-D groundwater flow and heat transport model (regional model) using FEFLOW was constructed in the Sendai Plain, Japan. Horizontal model domain was extended up to the water divide of the plain; its area is about 3600 km2. Vertical model domain covers from ground surface to the lower part of Neogene; the maximum model thickness is about 900 m. Calculated value of the simulation was verified based on field data of the groundwater levels and the subsurface temperature profiles. Next, identical ground heat exchanger models of dimensions 20 m × 20 m × 120 m were constructed at 33 locations to calculate heat exchange rates at these locations assuming a general closed-loop system based on the regional model. The operating scenario was set as 120 days of space heating per year from December to March assuming 24-h operation. The inlet temperature and flow rate of the circulation fluid were set as 5°C and 20 L/min, respectively. Heat exchange rate was calculated in the above condition and factors affecting the rate was examined by statistical methods. It was found that the main factors affecting heat exchange rate were groundwater flow velocity, subsurface temperature and thermal conductivity in the Sendai Plain. The equation using quadratic polynomial had better reproducibility of the amount of heat exchange than that using first-degree polynomial.