For the risk optimization of secondary distribution of refined oil, this study first selected the appropriate road risk model according to the public’s risk preference, and added road risk dynamic characteristics into the model to make it more fit the real situation. Next we introduced the concept of vulnerable center risk and constructed risk optimal model combined with the scale and the distance of vulnerable centers, which enriched the connotation of oil distribution risk. Then a multi-objective integer model including transportation costs and road risk and vulnerable center risk was formulated, the corresponding multi-layer search algorithm was also designed. Finally, the validity of this model and algorithm is verified through numerical experiment.