The objective of this paper is to compare the required length and performance of ground heat exchangers as well as heat pump energy consumption for fixed and variable speed ground-source heat pumps. In the first part of the paper, a physics-based model of a water-to-water heat pump is briefly presented. This model is incorporated in TRNSYS simulations using a performance map where variable speed operation is handled through a linear relationship linking the COP to the percentage of the full capacity being used. The ground heat exchanger is modeled using a thermal resistance and capacitance approach to account for borehole thermal capacity. Simulations are performed on a typical residential building located in a cold climate (Montreal, Canada) and equipped with either a fixed or variable speed ground-source heat pump. Results are obtained for eight cases with: variable or fixed speed operation (VSC or FSC), with or without consideration of borehole thermal capacity (TC or NTC), and with annual heating needs covered at 90% or 100% by the heat pump. The differences in the required borehole length between the TC and NTC cases are relatively small. The smallest required borehole length is for the FSC-90%-TC case (180 m) and the longest is for the VSC-100%-TC case (250 m). The VSC-100% case has the largest seasonal Performance factor (SPF) at 4.14 and the FSC-90% case has the lowest at 3.11.