A non-isothermal (linearly increasing temperature) procedure was used to determine the moisture diffusivity in mass transfer as a function of temperature with the complex optimization method. The effects of heating rate and experimental errors on the prediction accuracy of the parameters in Arrhenius equation were evaluated through pseudo-experimental data randomly generated on the basis of simulation results. An optimum heating rate of 0.01 K/s was found at which the experimental error less than 15% has negligible effect on calculated liquid diffusivity.