An experimental setup was built to obtain the experimental g-function of a small-scale borehole. The experimental g-function is calculated from the measured borehole wall temperature and the net heat injection rate into a 400mm long borehole inserted in a 2m3 sand tank with known thermal properties. With such a small length it is possible to reach a steady-state condition after a 168h long test. The experimental g-function of the borehole is compared to the g-function obtained from an analytical model based on the finite line source solution. The difference between the experimental and analytical g-functions is 4.7% after one week but the two curves remain within the experimental uncertainty bands. The air temperature variations during the test are shown to have a relatively significant effect on the results.