Results from the measurement of the heat of reaction of hydrothermal carbonization by power compensated differential scanning calorimetry exhibited a comparably high experimental standard deviation of around 10–20%. The reasons for this standard deviation have been investigated and are being presented in this article. The zeroline deviation and its repeatability showed a decisive influence on the measurements due to the length of the thermal effects (several hours) and the experimental setup (high thermal capacity due to pressure capsules and hydrothermal conditions, type of calorimeter). It was quantified by reference runs and compensated mathematically. In addition, conceptual issues due to the propagation of uncertainty by sum operations are derived. There is an optimum peak length after which the uncertainty rises due to this uncertainty propagation. This optimum is at a signal level within the noise level. However, the contribution of this uncertainty showed little significance compared to the zeroline deviation and thus could be neglected. Results from hydrothermal carbonization of glucose show a mean value of 1060 J/gdaf with a standard deviation of 14% for the presented experimental setup. These values include compensations of systematic errors, including the zeroline deviation, baseline correction, leakage, and transient effects, which are discussed in detail.