Purpose: Using finite element analysis (FEA), to determine the nominal shrinkage stress of a composite under different restriction conditions defined by the longitudinal compliance (LC) and C-factor (C) of the testing system, and by the elastic modulus of the bonding substrate (E). Materials and Methods: Eight axisymmetric models representing an experimental setup used to determine composite shrinkage stress were simulated. Composite thicknesses of 0.5 mm and 4 mm were tested, defining different C and volumes (C = 6 and vol = 14 mm3 or C = 0.8 and vol = 113 mm3, respectively). The E of the substrate was tested in two levels, 12 GPa and 207 GPa. Two LC values (1x10-6 or 28x10-6 mm/N) were defined for each E value by varying the length of the rods used as bonding substrate (0.3 mm and 9.5 mm for E = 12 GPa; 6.0 mm and 163.9 mm for E = 207 GPa). Materials were considered elastic, homogeneous, and isotropic. Shrinkage was simulated by thermal analogy. Nominal stress (nodal force/cross-sectional area) was calculated for each condition. Results were analyzed using Taguchi's method. Results: Nominal stress values varied between 1.7 MPa and 30.3 MPa. The main variables were statistically significant (LC: p = 0.0046; C: p = 0.0153; E: p = 0.0155), as well as the LC x E interaction (p = 0.0354). Stress reduction between low and high LC was more pronounced for E = 207 GPa compared to E = 12 GPa. Stress was lower for the high C conditions for both compliance levels. Conclusion: Not only the C-factor of the testing assembly, but also its LC and the E of the bonding substrate influence stresses generated by composite shrinkage. Keywords: C-factor, shrinkage stress, finite element analysis, compliance, elastic modulus