Abstract: The parameter regression for double-structural-parameters model of waxy crude oil was unstable. Based on the grouped progressive determination method of thixotropic parameters, the regression model of thixotropic parameters was established according to the principle of least square. Then, self-adaptive differential evolution algorithm with good stability was introduced to optimize the regression model. Furthermore, the initialization operation of this algorithm is improved to deal with the particularity of the problem. In the improved self-adaptive differential evolution algorithm, the phenomenon of random fluctuation occurs in the regression results of thixotropic parameters. Finally, a new experimental method of thixotropy measurement which combines step-shear rate experiment with constant-shear rate experiment was proposed based on the analysis of regression results. And thus, the instability of parameter regression results for double-structuralparameters model is effectively solved.