Laboratory-scale cone penetration tests are often carried out to calibrate the response of cone penetration, and in particular cone tip resistance to soil characteristics. However, because of the limited sample size, sample boundaries can often affect the measured cone resistance in laboratory tests. This paper presents numerical simulations using the discrete element method (DEM) to study the effect of boundary condition on cone penetration calibration chamber tests. Numerical simulations were performed under flexible (BC1) and laterally-constrained (BC3) boundary conditions on K0-consolidated models at different relative densities and vertical stresses. Additional models were simulated with periodic boundaries (BC5) to model the free-field condition. To track the radial stress variations in different sections of the chamber, scattered representative volume elements (RVE) were embedded in the models. Particle displacements and contact force chains were examined to determine the relation between microscopic variables and macroscopic response of the specimens subjected to cone penetration under different boundary conditions. Larger cone resistances were obtained under BC3 condition than those in BC1 condition. For the chamber-to-cone diameter ratio of 25 adopted in this study, the influence of the lateral boundary was found to be negligible in loose to medium-dense assemblies, while the effect of chamber boundary amplified in dense to very dense samples with increasing relative density and reduced with increasing vertical stress. This was attributed to the higher radial stress induced along the cone penetration path in laterally-constraint BC3 models. Based on these findings, a correction factor is proposed to better estimate free-field penetration resistance from calibration chamber experiments.