Objective: We aimed to perform a systematic review and meta-analysis to compare the treatment effects of unstable shoes and flat shoes on lower back pain patients. Data sources: Literature databases, including PubMed, Web of Science, and EMBASE (up to June 2019), were searched systematically. Review methods: Two authors independently screened the retrieved records and identified the randomized controlled trials where patients with lower back pain who wore unstable shoes as intervention and wore flat shoes as a control. Relevant data were extracted for meta-analysis using Review Manager 5.3 software. The Grading of Recommendations Assessment, Development and Evaluation approach was used to assess the pooled outcome evidence levels. Results: Five randomized controlled trials and 251 patients were included in the analysis. The meta-analysis results showed that there was a tendency toward a reduction in the Roland–Morris disability questionnaire score (mean difference (MD) –2.16, 95% confidence interval (CI) –4.28 to −0.03, I2 = 53%) and pain score (MD −0.84, 95% CI −1.66 to −0.02, I2 = 84%) in patients wearing unstable shoes compared to those wearing flat shoes. There was no significant difference in the life quality scores between the unstable shoe and flat shoe groups (MD −0.59, 95% CI −6.18 to 5.01, I2 = 0%). Functional disability and pain scores were determined to have very low-quality evidence, and life quality scores were determined to have low-quality evidence according to the Grading of Recommendations Assessment, Development and Evaluation analysis. Conclusion: Unstable shoes may be effective in treating lower back pain in the clinic, but the conclusion was limited by the current low-quality studies.