Heat transfer through can headspace was investigated using an inverse heat conduction problem (IHCP) approach. Solution for heat flux at the can headspace side was accomplished by using one dimensional heat conduction equation with one known boundary condition and internal temperature measurements. The estimation for the unknown heat flux at the headspace side was accomplished sequentially in time using the function specification algorithm. The accuracy of the estimated heat flux was investigated using the direct solution of the heat conduction equation employing the estimated heat flux to calculate the temperature distribution in the axial direction. From the knowledge of the calculated temperature at the sensor position and the experimental value the root mean square error (RMS) was calculated yielding a maximum value of 0.24 degrees C, this indicates the accuracy of the calculated heat flux. From the calculated heat flux and the surface temperature an effective heat transfer coefficients history was estimated for each of the three levels of headspace investigated (10%, 14% and 20% (v/v)). The calculated average effective heat transfer coefficients are 54.2, 53.2 and 50.8 W/(m(2)degrees C) for headspaces 10%, 14% and 20% (v/v), respectively. These values seem to be close due to presence of pure water vapor in the headspace with similar thermo-physical properties as dictated by phase equilibrium consideration resulting apparently in the same natural convection effects in the headspace. The technique used here could be useful for the estimation of an effective heat transfer coefficient for cases of water vapor gases mixture. (c) 2006 Elsevier Ltd. All rights reserved.