Complex regional pain syndrome (CRPS) is characterized by pain, limited range of motion, swelling, skin changes, vasomotor instability, and patchy bone demineralization. Conservative management strategies for CRPS include physical and occupational therapy, psychosocial and behavioral therapy, and pharmacotherapy. However, some patients still experience CRPS symptoms after receiving conventional treatments. Therefore, botulinum toxin (BoNT) has been applied to patients with CRPS in several trials considering its analgesic effect in musculoskeletal and neuropathic pain; however, the results were controversial. We conducted the study to explore the effectiveness and safety of BoNT in patients with complex regional pain syndrome (CRPS). A search was performed using the following electronic databases up to 19 October 2022: PubMed, Embase, and Cochrane Library. We included both randomized controlled trials and nonrandomized controlled studies involving patients with complex regional pain syndrome managed with botulinum toxin. Cochrane risk-of-bias tool and Joanna Briggs Institute Critical Appraisal Checklist were used for quality assessment for randomized controlled trials and quasi-experimental studies. Only randomized controlled trials entered the meta-analysis. The primary outcome was the visual analogue scale of pain presented as a weighted mean difference (WMD) and 95% confidence interval (CI). The secondary outcome was the risk of adverse events presented as an odds ratio (OR) with 95% CI. We analyzed eight articles with 176 patients, including three randomized controlled trials with 62 participants. The age of the patients ranged from 23.8 to 51 years old. The duration of the disease ranged from 2.2 to 11.8 years. The proportion of females ranged from 16.6% to 100%. The route of administration of BoNT included: (1) lumbar sympathetic block (LSB), (2) intramuscular injection, (3) subcutaneous or intradermal injection (SC/ID). Improvement in pain was revealed in six studies, and adverse events were all self-limited and temporary. Meta-analysis revealed a significant reduction in pain at the first follow-up between 3 weeks to 1 month after intervention (WMD, -1.036, 95% CI, -1.673 to -0.400) but not at the second follow-up between 2 to 3 months after treatment (WMD, -0.895, 95% CI, -2.249 to 0.458). Subgroup analyses between LSB and SC/ID were nonsignificant at both follow-up periods ( = 0.422, 0.139). The risk of adverse events was similar between the BoNT and control group (OR, 0.698, 95% CI, 0.136 to 3.581). In conclusion, BoNT may be effective and safe for alleviating pain in patients with CRPS. However, we could not draw definite conclusions due to small sample size and high between-study heterogeneity. The limited number of participants may conceal the possibility of serious adverse events. Further large-scale randomized controlled trials are warranted to delineate the role of BoNT in CRPS.