Abstract: With more than 90 years of development, the soil mechanics has become an indispensable theoretical analysis tool in geotechnical engineering. On the one hand, the soil mechanics is still a subject containing half theory and half experience, and its theoretical predictions are greatly uncertain. On the other hand, the geotechnical engineering problems become more complex and more important with the economic and social development. These situations force the development and reform of theories of soil mechanics. Firstly, the soil is demonstrated to be a kind of variable and sensitive material consisting of solid particles, pore water and pore air. Thus, it is difficult to accurately and quantitatively describe soil behaviors by using a simple one or two variables. Secondly, it is revealed that the principle of effective stress is a kind of simplified theory ignoring the effects of other factors on the strength and deformation of soils with the exception of effective stress. Therefore the classic soil mechanism is difficult to describe the complex soil behaviors, which results in a very large uncertainty of predicted results for soil behaviors. Thirdly, it is pointed out that the principle of effective stress is an equivalent approximation method, and the definition, expression and comprehension of the effective stress and its limitations are studied and discussed for saturated and unsaturated soils as follows: (1) In fact, the effective stress is an equation to describe the distribution of the total stress to the effective stress and the pore pressure. Then a two-phase saturated soil is simplified to a single-phase medium. The change of pore water pressure with void ratio, soil structure and other factors under given loading conditions implies that the effective stress is not an independent stated variable. (2) The single stress expressions for unsaturated soils is also proposed in the same way. However, its depicting ability for mechanical behaviors of unsaturated soils is not satisfactory. Then, the double stress variables are proposed to describe the mechanical behaviors of unsaturated soils. (3) It should be noted that the effective stress is the main controlling variable for the strength and deformation of saturated soils rather than the only one. Finally, in view of the current shortcomings of soil mechanics and the demand of geotechnical engineering problems, it is proposed to establish the soil mechanics considering the interaction and influence of multiple factors, and to advocate the study and description of soil behavior with the viewpoints of multiple factors (or multiple fields) rather than using only the effective stress.