Abstract: Objective Meta-analysis was used to evaluate the application of evidence-based nursing and the forecasting nursing in first aid for upper gastrointestinal bleeding. Methods The search databases included: China Knowledge Network, Wanfang Data, Weipu, China Biomedical Database, PubMed, MEDLINE, Cochrane and Embase, etc. We formulated searching key words and strategies. Literature quality assessment was performed using The Newcastle-Ottawa Scale (NOS). Data processing was conducted on RevMan 5.3 software, drawing a forest map. The effect index used RR value and 95%CI, and the heterogeneity test index was I² and P value. Sensitivity testing was performed and the results of the publication bias were evaluated using the Egger's method on the Stata 12.0 software. Results Compared with the predictive nursing group, Z=3.04 (P=0.002) and RR (95%CI): 1.05 (1.02, 1.08) in the evidence-based nursing group showed statistical significance (P<0.05). As for sensitivity analysis, after changing the effect model, eliminating low-quality literature, and eliminating case-control studies, the results were still statistically significant (P<0.05). As for the publication bias evaluation, t=0.97, P=0.343, 95%CI: (-0.62, 1.70) indicated no statistical significance (P>0.05). Conclusion Evidence-based nursing has a better effect on the hemorrhage of upper digestive tract than forecasting nursing, and it is worthy of clinical application.