Traditional anti-diabetic drugs may have negative or positive effects on risk of bone fractures. Yet the relationship between the new class glucagon-like peptide-1 receptor agonists (GLP-1 RA) and risk of bone fractures has not been established. We performed a meta-analysis including randomized controlled trials (RCT) to study the risk of bone fractures associated with liraglutide or exenatide, compared to placebo or other active drugs. We searched MEDLINE, EMBASE, and clinical trial registration websites for published or unpublished RCTs comparing the effects of liraglutide or exenatide with comparators. Only studies with disclosed bone fracture data were included. Separate pooled analysis was performed for liraglutide or exenatide, respectively, by calculating Mantel-Haenszel odds ratio (MH-OR). 16 RCTs were identified including a total of 11,206 patients. Liraglutide treatment was associated with a significant reduced risk of incident bone fractures (MH-OR=0.38, 95% CI 0.17-0.87); however, exenatide treatment was associated with an elevated risk of incident bone fractures (MH-OR=2.09, 95% CI 1.03-4.21). Publication bias and heterogeneity between studies were not observed. Our study demonstrated a divergent risk of bone fractures associated with different GLP-1 RA treatments. The current findings need to be confirmed by future well-designed prospective or RCT studies.