Two full-scale composite floor slabs were tested at Ferguson Structural Engineering Laboratory at The University of Texas at Austin under two different column removal scenarios. The removal of a column and the associated response of a structure is an index of its resiliency under abnormal loads, such as those due to a terrorist attack or a vehicle collision. Previous computational studies have shown that floor slab contributions are extremely important in mitigating collapse, but the limited experimental data currently available provide inconsistent results. The aim of the experimental testing program was to identify basic behaviors of floor slabs and to estimate their ultimate capacity under the absence of a critical column. The two test specimens were representative of isolated sections of the gravity-load resisting system of a typical steel-framed building. Thus, all steel members were joined using simple connections. During testing, the critical column was statically removed under service loads. Next, the load on the floor slab was increased at a slow rate until the specimens completely collapsed. Overall, the ultimate load carrying capacity of the two specimens under the absence of a single column exceeded the required capacity from progressive collapse provisions. Detailed finite element models were developed and validated against the collected experimental data in which all the components of the floor system were explicitly modeled. The explicit nonlinear finite element software LS-DYNA® was employed to simulate the response of the experimental tests. Initially, individual components of the floor system were modeled and validated against experimental data available in the literature. The two specimens were modeled using a similar approach. The main components of the floor system were modeled using three-dimensional solid elements for the concrete and steel members, shell elements for the corrugated steel deck, and beam elements for the shear studs and reinforcement in the slab. Bolts and other connection components were explicitly modeled using solid elements, and contact was specified to account for the interaction among the connected parts. Good agreement was found between the tests and numerical simulations. Further analyses provided information about the sensitivity of the numerical models to several design parameters.