This study aimed to investigate the effects of preliminary traction on the rate of failure of reduction and the incidence of femoral head avascular necrosis (AVN) in patients with late-detected developmental dysplasia of the hip treated by closed reduction. A total of 385 patients (440 hips) treated by closed reduction satisfied the inclusion criteria. Patients were divided in two groups according to treatment modality: a traction group (276 patients) and a no-traction group (109 patients). Tönnis grade, rate of failure reduction, AVN rate, acetabular index, center-edge angle of Wiberg, and Severin's radiographic grade were assessed on plain radiographs, and the results were compared between the two groups of patients. In addition, a meta-analysis was performed based on the existing comparative studies to further evaluate the effect of traction on the incidence of AVN. Tönnis grade in the traction group was significantly higher than in the no-traction group (P = 0.021). The overall rate of failure reduction was 8.2%; no significant difference was found between the traction (9.2%) and no-traction groups (5.6%) (P = 0.203). The rates of failure reduction were similar in all Tönnis grades, regardless of treatment modality (P > 0.05). The rate of AVN in the traction group (14%) was similar to that of the no-traction group (14.5%; P = 0.881). Moreover, the rates of AVN were similar in all Tönnis grades, regardless of treatment modality (P > 0.05). The meta-analysis did not identify any significant difference in the AVN rate whether preliminary traction was used or not (odds ratio = 0.76, P = 0.32). At the last follow-up visit, the two groups of patients had comparable acetabular indices, center-edge angles, and Severin's radiographic grades (P > 0.05). In conclusion, preliminary traction does not decrease the failure of reduction and the incidence of AVN in developmental dysplasia of the hip treated by closed reduction between 6 and 24 months of age.