The demands on machining accuracy have been increasing lately and therefore research of thermal behavior of machine tool structures is crucial for successful manufacturing. Generated heat diffuses into the structure of the machine tool components, this process is affected by heat sinks such as heat transfer on the surfaces and cooling systems. Meanwhile the heat warms up the structure of the machine tool and thermal dilatation deforms the structure, which subsequently affects machining accuracy in a negative way. Different systems are used to eliminate the thermal error, but their efficiency corresponds to the quality of the thermal machine tool model. The key problems of machine tool thermal error reduction are not in the thermal model itself, but in the fast, or even real-time, identification of the heat sources and the heat transfer coefficients on the surfaces. The identification process represents inverse heat conduction problem (IHCP), where the temperatures of the structure are the input parameters. This paper brings up a new solution of such IHCP, which enables a real-time identification of the heat transfer coefficient. The solution is founded on an analytical description of the heat transfer in a thin plate. The identification method is tested on a finite element model and experimentally.