This paper presents a method for the calibration of contact heat flow sensors and the measures taken to improve the overall standard uncertainty of the measurements. This may be particularly relevant for applications such as the thermal characterization of Li-ion battery cells, where small amounts of transferred heat must be measured. The designed calibration set-up is based on the measurement of temperature gradients within a reference object, which is a common and well known approach. However, certain aspects must be considered in order to obtain accurate results. In this paper all steps taken to improve the overall measurement uncertainty are described. One of the most important characteristics of the presented methodology is the experimental determination of an equivalent thermal conductivity of the used reference object. To do so, a single-sided guarded hot plate apparatus is integrated into the calibration set-up to generate a precisely known heat flow. This approach is comparable to the determination of the ‘energy equivalent’ in calorimeters. The measured equivalent thermal conductivity agrees well to literature values, but the achieved standard uncertainty is significantly lower. By way of example, the calibration method is applied on a thermoelectric generator, where the temperature dependent sensitivity and the Seebeck coefficient are determined and compared to literature values. The average standard uncertainty of the sensor sensitivity is ±1.86%, which is lower than most results published in the literature. At 20 °C the absolute value of the determined Seebeck coefficient of the examined thermoelectric element is only 2.7% higher than the value expected for the used material (
B
i
2
T
e
3
).