Aiming at the serious shortage of relief in the early stage after the earthquake, a bi-level collaborative optimization model of relief allocation-transportation is formulated based on fairness and efficiency by considering the non-rational psychological comparison of the victims. The top level model guarantees the fairness of relief allocation by minimizing the sum of loss comparison effect of all the demand points. To guarantees the fairness and efficiency of relief transportation, the bottom level model minimizes the sum of time comparison effect of relief reaching demand points, and minimizes the total time of relief transportation. The proposed model is solved by a hybrid genetic algorithm. Finally, the numerical experiments illustrate the rationality and effectiveness of the proposed approaches.