To prevent the outflow of nourished sand on the landside, two types of sand fence arrangements were tested on a natural beach, namely (1) a configuration that combined a straight and a transverse alignment, and (2) a configuration that combined a straight and a perpendicular alignment. The former configuration was superior in its function, although the latter configuration also worked satisfactorily. In addition, long‐term field observations of the beach evolution during 15 years in response to a particular sand fence arrangement demonstrated that two rows of sand fences parallel to the shore, auxiliary employed on the landward side, also functioned well. The sand volume that each configuration can control was evaluated by assuming a suitable accumulation form around the sand fences. The configuration that combines a straight and a transverse alignment will effectively trap sand with under the assumption that the transverse alignment can form a wide dune parallel to the shoreline. Evaluation of the configuration that combined a straight and a perpendicular alignment could not be performed because sand accumulation forms around fences were too complicated and did not adhere to a suitable accumulation form. For the two rows of sand fences, a scalene triangle for the accumulation form was applied to the fences having low porosities located at the seaside. It was concluded that the evaluation could be performed with satisfactory accuracy. Then, based on the results obtained from these experiments and previous knowledge, suitable sand fence configurations for winds blowing obliquely to the shore, as well as for winds blowing normally to the shore, were proposed together with guidelines to determine dimensions of each element composing the fence configuration.