The few heat flow measurements that exist mean that each measurement becomes a regional evaluation. However, heat flow depends largely on the geological context. In this paper, we study configurations which may disturb heat flow density distribution : thermal conductivity contrast between two areas, free or mixed convection cells, forced convection in a dipping layer, hot springs deriving from bedrock. As a case study, a geothermal anomaly which has been observed near Vittel (Vosges, France) is analyzed according to the principles outlined above. The only hypothesis consistent with the observations is that hot springs occurring along the Vittel fault disturb the normal regional heat flow which is evaluated at being about 80 mWm^-2. Conclusions concerning the representativeness of heat flow measurements are drawn up.