A set of standard quenching tests[1] was performed in order to assess the effect of different variables on heat transfer coefficient (HTC). Different levels of oil temperature, initial probe temperature and agitation were considered. A novel filter technique to post process the experimental data and an algorithm to identify the HTC were proposed. The filter (based on non-linear diffusion technique) generates smooth cooling rate curves needed for further HTC determination. The resolution of the inverse thermal problem was based on the iterative resolution of the corresponding direct problem up to convergence of the flux boundary condition[2]. The method uses an appropriate seed function of heat transfer coefficient which is applied as boundary condition and corrected after every iteration up to convergence of the numerical cooling rate curve (t,dT/dt). As a result, HTC's curves having high level of detail and smoothness were obtained. The obtained results were used for the identification of the different heat transfer mechanisms that take place during quenching. The generation of a parametric heat transfer model based on mean test conditions is discussed.