We investigate the problem of artifacts caused by the periodic boundary conditions (PBC) used in molecular simulation studies. Despite the long history of simulations with PBCs, the existence of measurable artifacts originating from PBCs applied to inherently nonperiodic physical systems remains controversial. Specifically, these artifacts appear as differences between simulations of the same system but with different simulation-cell sizes. Earlier studies have implied that, even in the simple case of a small model peptide in water, sampling inefficiency is a major obstacle to understanding these artifacts. In this study, we have resolved the sampling issue using the replica exchange molecular dynamics (REMD) enhanced-sampling method to explore PBC artifacts. Explicitly solvated zwitterionic polyalanine octapeptides with three different cubic-cells, having dimensions of L = 30, 40, and 50 Å, were investigated to elucidate the differences with 64 replica × 500 ns REMD simulations using the AMBER parm99SB force field. The differences among them were not large overall, and the results for the L = 30 and 40 Å simulations in the conformational free energy landscape were found to be very similar at room temperature. However, a small but statistically significant difference was seen for L = 50 Å. We observed that extended conformations were slightly overstabilized in the smaller systems. The origin of these artifacts is discussed by comparison to an electrostatic calculation method without PBCs.