ABSTRACT We report three-dimensional hydrodynamical simulations of shocks (${\cal M_{\rm shock}}\ge 4$) interacting with fractal multicloud layers. The evolution of shock–multicloud systems consists of four stages: a shock-splitting phase in which reflected and refracted shocks are generated, a compression phase in which the forward shock compresses cloud material, an expansion phase triggered by internal heating and shock re-acceleration, and a mixing phase in which shear instabilities generate turbulence. We compare multicloud layers with narrow ($σ _{ρ }=1.9\bar{\rho }$) and wide ($σ _{ρ }=5.9\bar{\rho }$) lognormal density distributions characteristic of Mach ≈ 5 supersonic turbulence driven by solenoidal and compressive modes. Our simulations show that outflowing cloud material contains imprints of the density structure of their native environments. The dynamics and disruption of multicloud systems depend on the porosity and the number of cloudlets in the layers. ‘Solenoidal’ layers mix less, generate less turbulence, accelerate faster, and form a more coherent mixed-gas shell than the more porous ‘compressive’ layers. Similarly, multicloud systems with more cloudlets quench mixing via a shielding effect and enhance momentum transfer. Mass loading of diffuse mixed gas is efficient in all models, but direct dense gas entrainment is highly inefficient. Dense gas only survives in compressive clouds, but has low speeds. If normalized with respect to the shock-passage time, the evolution shows invariance for shock Mach numbers ≥10 and different cloud-generating seeds, and slightly weaker scaling for lower Mach numbers and thinner cloud layers. Multicloud systems also have better convergence properties than single-cloud systems, with a resolution of eight cells per cloud radius being sufficient to capture their overall dynamics.