In the self-balanced loading test, reasonable determination of conversion factor γ, ratio of downward shaft resistance to upward shaft resistance, is critically important for the accuracy of bearing capacity and final decision of pile loading test. To study the appropriate value of γ in sandy soil, 3 typical construction projects are chosen. All of their main soil layer are sandy soil; and the self-balanced loading test and head-down loading test are conducted at every site. And then a pile from self-balanced loading test and one from head-down loading test are picked as a group. In these groups, the geometrical sizes, geological condition and construction method of the two piles are same or similar. Because the results from traditional loading test are more convincing than ones from self-balanced loading test in general, the results from top-down loading test are taken as references; and then the Matlab is used to find the optimal fitting solution of γ, latterly, returning the value to the simplified capacity equation of the self-balanced loading test, and comparing the result of equivalent top-loaded settlement curve with ones from γ used in actual project and from head-down loading test, ? = 0.6 and 0.7. The results show that: (1) The fitted values of ? from 3 projects are ranged from 0.47 to 0.71, which are rather smaller than that used in actual projects. (2) The accuracy of ? has large effect on the values of Qs (skin friction) and sss (elastic compression from skin friction distribution) of upper piles of the equivalent pile. (3) Compared with the top-loaded settlement curve from traditional loading test, the results from the optimal fitting solution of ? fit much better than of which used in actual project ,which means that values of ? used in these projects are overly conservative, which cause overly conservative judgement of bearing capacity. (4) It is suggested that the value of conversion factor ? in sandy soil about 0.6 to 0.7.