PURPOSE: To investigate the etiologic factors related to refractures of the upper extremity in children MATERIALS AND METHODS: 18 refractures of the upper extremity were divided into three groups according to the location of initial fractures: Supracondyle fractures of the humerus, lateral condyle fracture of the humerus, and the forearm bone fractures. They were analyzed in terms of the type of refractures (early refracture occurring at the immature callus and late refracture occurring at the remodeled bone), fracture patterns, and the existence of underlying deformity. RESULTS: Nine supracondyle fractures had refractures at the supracondyle (2 cases) and the lateral condyle (7 cases), in which underlying cubitus varus were present in 6 cases. Three lateral condyle fractures had refractures at the supracondyle (1 case) and the lateral condyle (2 cases), in which one case had underlying cubitus varus. All but one case in the group of humerus fractures were late refractures and treated operatively except one. Of 6 refractures of forearm, 5 were early refractures and occurred within 9 weeks at the original site: 4 at the diaphysis of both bones of forearm and 1 at the diaphysis of ulna. All cases in the group of forearm fractures had volar angulation before the refracture, and treated conservatively except one CONCLUSION: In the humerus, underlying cubitus varus was the most important predisposing factor to refractures and the lateral condyle fractures were common. In the forearm, volar angulation of the diaphysis were related to refractures, and complete and circular consolidation of the primary fracture of forearm was thought to be important to prevent refracture.