Abstract While clinical reports demonstrate that CTLA-4 Ig (abatacept) is efficacious in rheumatoid arthritis (RA) patients refractory to standard therapies, its mode of action is poorly characterized. Using the Entelos® RA PhysioLab® platform, a mechanistic mathematical model of a prototypical human RA joint, we evaluated abatacept’s potential mechanisms of action to gain insight into the basis of its clinical efficacy. The hypothesized effects of abatacept on (a) T cell functions (e.g., inhibition of proliferation, costimulation, and Th1 activity and induction of T regulatory cells (Tregs)), (b) antigen presenting cell functions (e.g., inhibition of cytokine production and antigen presentation), and (c) cross talk between T cells and macrophages, fibroblast-like synoviocytes, and dendritic cells were investigated by altering the model values that define each process. Biosimulation of abatacept therapy demonstrated that a 50% modulation of these targeted functions is sufficient to reproduce clinical responses in a ‘virtual’ RA patient, whereas a 20% effect is insufficient. Analysis of these results shows that inhibition of macrophage activation and cytokine production, and skewing of T cells to a Treg phenotype drive abatacept efficacy in silico. By reducing the uncertainty regarding the mechanisms critical for abatacept efficacy, biosimulation can lead to better and more targeted RA treatments.