<p indent="0mm">The accurate and comprehensive understanding of the molecular composition, source, and stabilization mechanism of soil organic matter (SOM) is vital for soil management and the preservation of organic carbon. Currently, a series of key scientific issues concerning the formation and stabilization of SOM is highly contentious. This review summarized the advances in the sources and molecular structure of SOM, and presented a systematic analysis of the different views, conflicts, and problems concerning soil humic substances in the past decades. The traditional view of soil science believes that SOMs are mainly composed of humic substances which are macromolecular compounds and are derived from plant and animal residues through a series of mineralization and humification processes. Soil enzymes and minerals, including phenol oxidase, laccase, peroxidase, iron oxide, manganese oxide, promote the polymerization of organic molecules through additive nucleophilic reactions and free radical catalyzation. <sup>13</sup>C-NMR and FT-ICR-MS analyses supported the existence of humic substances in soils based on the structure and the molecular composition of SOM. However, with the application of biomarker analysis, high-resolution mass spectrometry, spatial resolution spectroscopy, and other advanced techniques, there is an emergence of new perspectives on the source and the structure of SOM. HPLC-SEC showed the average molecular weight of Aldrich humic acid was significantly smaller than previously thought and the aggregation of humic substances was affected by acetic acid, pH and ionic strength. These evidences suggest that humic substances are aggregates of small organic compounds with a molecular weight of hundreds to thousands of Daltons, rather than macromolecules of tens to hundreds of thousands of Daltons. Furthermore, <italic>in situ</italic> observation of soil particles by STXM revealed significant heterogeneity of SOM at <sc>50 nm</sc> spatial resolution. In terms of the origin of SOM, model calculations and biomarker analysis have indicated that microbial necromass is the major constituent of SOM which accounts for ~33%–62% of SOM in different ecosystems. Collectively, microbial residues may contribute predominantly to the SOM pool and traditional concept “humification” may not occur in the soil. These new findings are contrary to the traditional theory of soil chemistry and soil biochemistry and subvert our understanding of SOM. Therefore, it is imperative to perform further in-depth studies to fill the knowledge gaps regarding the sources, formation processes, and molecular structures of humus. However, further intensive investigations on SOM are restricted by the lack of efficient and reliable extraction procedures. In addition, the conversion coefficients from biomarker to necromass are not accurate and mass spectrometry could not provide a comprehensive analysis on the whole organic molecules in the soil. The studies on enzymatic reactions and microbial synthesis were mainly performed in simple suspension systems, which could not simulate the complex soil matrix. Consequently, future studies are needed to improve the efficiency and accuracy of the extraction and analysis of SOM and to develop a real-time study platform for the interfacial reactions among minerals, organic matters and microorganisms. Meanwhile, <italic>in situ</italic> characterization techniques and modeling should be employed and combined to clarify the sources and the underlying mechanisms of SOM stabilization at molecular level.