This paper proposes a method to estimate in-soil deformation behavior of geogrid based on the results of direct box shear tests. A series of drained direct box shear (DBS) tests were performed, consisting of two types of normal stress loading methods and a series of pull-out tests in which the space of the pull-out opening, embedded length of geogrid and conditions of in-soil end restraint were changed. Comparisons were made between the calculated in-soil deformation behavior of the geogrid based on the results of the DBS tests and that of the results observed by pull-out tests. Based on this study, the following findings were obtained. Strength parameters corresponding to the peak or the residual states obtained from the two types of direct box shear tests almost coincide with each other. Using the method proposed in this paper with the strength parameters obtained by DBS tests, it is possible to estimate the in-soil behavior of geogrid. As the pull-out force vs. pull-out displacement relationships depend not on the conditions of in-soil end restraint of geogrid but on the normal stress exerted on the geogrid, it is important to construct a reinforced soil structure so that sufficient frictional resistance develops between soil and geogrid.