This work shows that toughness (resistance) to fracture propagation is an inherent characteristic of cohesionless particulate materials, which is significant for understanding hydraulic fracturing in geotechnical, geological, and petroleum applications. We have developed experimental techniques to quantify the initiation and propagation of fluid-driven fractures in saturated particulate materials. The fracturing liquid is injected into particulate materials, where the fluid flow is localized in thin crack-like conduits. By analogy, we call them 'cracks' or 'hydraulic fractures'. Based on the laboratory observations and scale analysis, this work offers physical concepts to explain the observed phenomena. When a fracture propagates in a solid, new surfaces are created by breaking material bonds. Consequently, the material is in tension at the fracture tip. In contrast, all parts of the cohesionless particulate material (including the tip zone of hydraulic fracture) are likely to be in compression. In solid materials, the fluid front lags behind the front of the propagating fracture, while the lag zone is absent for fluid-driven fractures in cohesionless materials. The compressive stress state and the absence of the fluid lag are important characteristics of hydraulic fracturing in particulate materials with low, or no, cohesion. Our experimental results show that the primary factor affecting peak (initiation) pressure is the magnitude of the remote stresses. The morphology of fracture and fluid leak-off zone, however, changes significantly not only with stresses, but also with other parameters such as flow rate, fluid rheology, and permeability. Typical features of the observed fractures are multiple off-shots and the bluntness of the fracture tip. This suggests the importance of inelastic deformation in the process of fracture propagation in cohesionless materials. Similar to solid materials, fractures propagated perpendicular to the least compressive stress. However, peak injection pressures are significantly greater than the maximum principle stresses in the experiments. Further, by incorporating the dominate experimental parameters into dimensionless form; a reasonable power-law fit is achieved between a dimensionless peak injection pressure and dimensionless stress. Scaling indicates that there is a high pressure gradient in the leak-off zone in the direction normal to the fracture. Fluid pressure does not decrease considerably along the fracture, however, due to the relatively wide fracture aperture. This suggests that hydraulic fractures in unconsolidated materials propagate within the toughness-dominated regime. Furthermore, the theoretical model of toughness-dominated hydraulic fracturing can be matched to the experimental pressure-time dependences with only one fitting parameter. Scale analysis shows that large apertures at the fracture tip correspond to relatively large 'effective' fracture (surface) energy, which can be orders of magnitude greater than typical for hard rocks.