Starting with the standard (i.e. linear) pre-dose technique, thermoluminescence dating of porcelain was handled. In the study the nonlinear response of sensitivity to dose and an exponential model in sensitivity change with dose were found, therefore a regression method of saturation exponential for evaluating the paleodose was proposed. It has been proved that this method is successful for thermoluminescence dating of porcelain. In addition, some complicated factors associated with thermoluminescence dating were discussed. For example, when sensitivity is not an exponential function of dose in a few porcelains, the regression model of paleodose must be changed, or an average value B is used.