Background: Although a variety of targets for deep brain stimulation (DBS) have been found to be effective in Parkinson's disease (PD), it remains unclear which target for DBS leads to the best improvement in gait disorders in patients with PD. The purpose of this network meta-analysis (NMA) is to compare the efficacy of subthalamic nucleus (STN)-DBS, internal globus pallidus (GPi)-DBS, and pedunculopontine nucleus (PPN)-DBS, in improving gait disorders in patients with PD.Methods: We searched the PubMed database for articles published from January 1990 to December 2020. We used various languages to search for relevant documents to reduce language bias. A Bayesian NMA and systematic review of randomized and non-randomized controlled trials were conducted to explore the effects of different targets for DBS on gait damage.Result: In the 34 included studies, 538 patients with PD met the inclusion criteria. The NMA results of the effect of the DBS “on and off” on the mean change of the gait of the patients in medication-off show that GPi-DBS, STN-DBS, and PPN-DBS are significantly better than the baseline [GPi-DBS: –0.79(–1.2, –0.41), STN-DBS: –0.97(–1.1, –0.81), and PPN-DBS: –0.56(–1.1, –0.021)]. According to the surface under the cumulative ranking (SUCRA) score, the STN-DBS (SUCRA = 74.15%) ranked first, followed by the GPi-DBS (SUCRA = 48.30%), and the PPN-DBS (SUCRA = 27.20%) ranked last. The NMA results of the effect of the DBS “on and off” on the mean change of the gait of the patients in medication-on show that, compared with baseline, GPi-DBS and STN-DBS proved to be significantly effective [GPi-DBS: –0.53 (–1.0, –0.088) and STN-DBS: –0.47(–0.66, –0.29)]. The GPi-DBS ranked first (SUCRA = 59.00%), followed by STN-DBS(SUCRA = 51.70%), and PPN-DBS(SUCRA = 35.93%) ranked last.Conclusion: The meta-analysis results show that both the STN-DBS and GPi-DBS can affect certain aspects of PD gait disorder.