The principles of the zero‐balance heat‐flux meter (HFM) technique are studied with two steady‐state finite difference computer models developed to simulate the responses of a sensor mounted either on the cold face or on the warm face of a wall. Computations show that a particular Biot number (defined by the authors), the dimensionless wall width, and the dimensionless wall thickness are the parameters affecting the HFM accuracy. The authors present outputs of their models that show the influence of these parameters, and propose three empirical expressions that allow corrections of the operational errors. It can be affirmed that the flux meter should be mounted on the cold face in the case of a small Biot number (it assures an accuracy of at least 5% if the Biot number value is smaller than 0.2). And it should be mounted on the warm face when the Biot number is large (in the case of building insulators in normal environment conditions it assures an accuracy of at least 10%). In other cases, the flux meter should be used as a gradient flux meter switching off the internal heater.