The wetting properties of a fluid in contact with an attractive substrate depend significantly on the properties of that substrate, including its roughness. This relationship has helped scientists create novel materials with applications ranging from superhydrophillic surfaces used in film spreading to biomimetic self-cleaning materials. In this presentation we describe our recent computational efforts aimed at understanding the link between wetting properties and the molecular-level roughness of a surface. Grand canonical transition matrix Monte Carlo simulations are used to obtain the surface free energy of a fluid in contact with an atomistically detailed surface. This information is subsequently used to calculate spreading coefficients at conditions below the wetting temperature and prewetting saturation points in the partially wetting regime. Extrapolations of these properties allow us to pinpoint the temperature of the wetting transition. To study wetting on rough surfaces, we use a Lennard-Jones fluid interacting with a substrate comprised of static particles. A collection of substrates with varying degrees of molecular disorder are considered, including various orientations of the perfect face-centered cubic lattice, thermally disordered crystalline configurations, and amorphous structures. A Voronoi tessellation is used to characterize the surface topology of a given structure, which provides a means to quantify the molecular roughness of the substrate.