Background: Insoles have been widely applied to many diseases, but stroke involves complex problems and there is a paucity of research on the application of insoles in stroke patients. Aim: To evaluate the effect of customized insoles on gait in patients with hemiplegia. Design: A randomized controlled trial. Setting: Rehabilitation department of a hospital. Population: A total of 50 stroke patients were randomized into an experimental group (n = 25) or a control group (n = 25). Methods: Both groups received conventional gait training, which was conducted five times a week, every 40 min for four weeks and patients in the experimental group were required to wear customized insoles for at least 1 h per day for four weeks. The primary outcome measure was the Tinetti Gait Scale (TGS) and the secondary outcome measures were the plantar pressure test, 6-min walking test (6MWT), lower extremity Fugl–Meyer assessment (FMA-LE), Berg Balance Scale (BBS), and the modified Barthel index (MBI). Results: Compared to the control group, there were significant increases in the experimental group after four weeks (p = 0.014) and at the four week follow-up (p = 0.001) in the change in TGS, weight-bearing on the involved side (p = 0.012) or forefoot (p = 0.028) when standing, weight-bearing on the involved side (p = 0.01 6) or forefoot (p = 0.043) when walking, early stance phase (p = 0.023) and mid stance phase (p = 0.013) on the involved side, FMA-LE (p = 0.029), BBS (p = 0.005), and MBI (p = 0.009), but there were no differences in the late stance phase (p = 0.472) on the involved side when walking or in the 6MWT (p = 0.069). Conclusions: Customized insoles had great efficacy in enhancing gait performance in stroke patients.