Background The exponential growth of high-throughput sequencing technologies was an incredible opportunity for researchers to combine various -omics within computational frameworks. Among these, metagenomics and metabolomics data have gained an increasing interest due to their involvement in many complex diseases. However, currently, no standard seems to emerge for jointly integrating both microbiome and metabolome datasets within statistical models. Results Thus, in this paper we comprehensively benchmarked nineteen different integrative methods to untangle the complex relationships between microorganisms and metabolites. Methods evaluated in this paper cover most of the researcher's goals such as global associations, data summarization, individual associations, and feature selection. Through an extensive and realistic simulation we identified best methods across questions commonly encountered by researchers. We applied the most promising methods in an application to real gut microbial datasets, unraveling complementary biological processes involved between the two omics. We also provided practical guidelines for practitioners tailored to specific scientific questions and data types. Conclusion In summary, our work paves the way toward establishing research standards when mutually analyzing metagenomics and metabolomics data, building foundations for future methodological developments.