Ethnopharmacological relevance: Network pharmacology is extremely adaptive for investigating traditional ethnic drugs, especially the herbal medicines. However, challenges still hang over many related studies due to the limitations in the methodology of conventional network pharmacology. Aim of the study: Our work was aimed to investigate the methodology limitations of conventional network pharmacology with Xian-Ling-Gu-Bao (XLGB) as a representative, meanwhile, propose the strategies for coping with these issues. Materials and methods: Predicted phytochemical constituents formed virtual XLGB. The constituents in realistic XLGB samples was detected by liquid chromatography-mass spectrometry (LC-MS) to correct the constituent deviation resulted from virtual prediction. Multivariate statistical analysis of quantitative target data were used to reveal the relation of target profile between drug and disease. The key constituents and targets were screened and compared between virtual and realistic XLGB through network analysis. After enrichment analysis, reversing network pharmacology was performed to exclude weak targets and re-construct the interaction from key pathways to key targets. Finally, the core constituents and action mechanism of XLGB were deduced. Results: Significant deviation of phytochemical constituents was found between virtual and realistic XLGB. As expected, this deviation led to a cascade of deviation ranging from deduced key constituents to key targets and key pathways. Moreover, many key KEGG pathways were enriched and screened out, however, they were almost irrelevant to the studied disease. These results systemically illustrated the limitations in the methodology of conventional network pharmacology. Importantly, the strategies for coping with these limitations were proposed, such as high-throughput detection of the realistic samples, multivariate analysis of target profile and combined enrichment analysis. Finally, based on the improved network pharmacology, the medicinal constituents and mechanism of XLGB against osteoarthritis were effectively deduced. Conclusions: Our work highlighted the necessity and proposed the strategies for improving the methodology of conventional network pharmacology. The corrected results from improved network pharmacology provided promising directions for future research on XLGB.