This thesis investigates the stability of coal pillars under realistic conditions of varying weak floor thickness with and without the use of paste backfill. Weak floor strata underlying coal seams are common in the Illinois Basin. They consist mainly of underclay, which is a gray, argillaceous rock that usually occurs immediately beneath beds of coal. Underclay thickness may vary from less than a foot to twenty feet at different locations in the basin (Grim and Allen, 1938). Locally, underclay thickness may vary gradationally over a distance of two pillars. Even though weak floor thickness is not consistent (Gadde, 2009), most research to date has focused on parametric studies with a fixed underclay thickness and formulated coal pillar designs on the basis of the maximum underclay thickness measured in the field. Therefore, it is necessary to investigate more realistic field conditions and quantify the influence of a gradated weak floor thickness using additional parametric studies. This research is primarily numerical modeling incorporating various constitutive models and using some calibration. Therefore, the two dimensional plane strain finite difference model in FLAC 3D is employed to carry out parametric studies on gradated weak floor conditions. Underclay exhibits Mohr Coulomb elastic plastic behavior; hence, the Mohr Coulomb constitutive model is used for the behavior of overburden, coal, and floor. Well-calibrated numerical models can assist in understanding load and failure processes provided that coal, overburden, and weak floor are modeled with sufficient realism. The theoretical approach considers a friction angle of 0° to calculate the load bearing capacity of the weak floor for design of pillars with long-term stability, even if the weak floor has a non-zero friction angle. The stiffness of the weak floor increases with an increase in friction angle (Gadde, 2009; Kostecki and Spearing, 2015). As stiffness increases, a point can be reached where floor bearing capacity exceeds coal pillar strength and coal pillar strength becomes the governing factor. For this scenario, the Mohr Coulomb strain softening model is more realistic in estimating loads carried by coal pillars in the post-failure stage. The three-dimensional Mohr Coulomb strain softening model in FLAC 3D is employed to study qualitatively the floor response in strain softening coal behavior conditions. Maintaining stable coal pillar responses has been a challenge for the coal mining industry due to attempts to increase the primary extraction ratio. Presently, the best available solution seems to be backfilling when considering short-term pillar stability (i.e., less than the long-term factor of safety) with increased extraction ratio. There are various types of mine backfill that have benefits to the mining industry depending on the application, but paste backfill produced from total mill tailings containing no free water is the best option for post-mining ground control in room-and-pillar mines as it prevents weakening of the floor and will not contaminate the ground water. The influence of paste backfill on floor bearing capacity and coal pillar response is studied with numerical modeling using the same constitutive models already identified. Finally, an economic analysis is carried out to look at cost implications of a proposed system with backfill.