Emerging atomically thin nanomaterials such as two-dimensional (2-D) graphene have attracted tremendous attention for their many unique properties. However, a single piece of them is too delicate to be useful in most applications, for example, high-performance electrodes in energy storage, filters for waste water/gas treatments in environmental systems, and lightweight structures. Assembling these nanomaterials into three-dimensional (3-D) scaffolds to achieve superior overall performance with multiple functionalities has attracted growing interests, yet this is challenging in manufacturing. In particular, these low-dimensional nanomaterials tend to aggregate/restack due to strong van der Waals attraction between them such as restacking of 2-D flat graphene sheets, which not only results in a tremendous reduction of their accessible surface area and poor mass/ion transport, but also degrades with processing and/or application environments such as mechanical loadings, hence adversely affecting their properties and subsequent applications. A liquid evaporation-assisted manufacturing technique is considered to provide a facile route, where 2-D nanomaterials will experience large deformation and severe instability under evaporation-induced compression to create spacings when assembled, which is highly desirable to minimize restacking and retain the large surface areas of 2-D nanomaterials in the assembled 3-D architectural structures. In this study, we establish a theoretical mechanics framework to quantitatively describe the liquid evaporation-driven deformation and self-assembly of 2-D graphene suspended in a liquid environment. The energy competition among surface energy of liquid, solid-liquid interfacial energy, solid-solid interactive energy, and deformation energy of solids during liquid evaporation is probed and incorporated into the mechanics theory. The critical deformation lengths of graphene sheets, and size and configuration of ultimate stable assembled 3-D particles are predicted and validated with extensive molecular dynamics simulations.