Manual wheelchair (MWC) locomotion exposes the user's upper-body to large and repetitive loads, which can lead to upper limbs pain and injuries. A thinner understanding of the influence of MWC settings on propulsion biomechanics could allow for a better adaptation of MWC configuration to the user, thus limiting the risk of developing such injuries. Advantageously compared to experimental studies, simulation methods allow numerous configurations to be tested. Recent studies have developed predictive locomotion simulation using optimal control methods. However, those models do not consider MWC anteroposterior stability, potentially resulting in unreasonable propulsion strategies. To this extent, this study aimed at confirming if constraining MWC anteroposterior stability in the optimal control formulation could lead to a different simulated movement. For this purpose, a four-link rigid-body system was used in a forward dynamics optimization paired with an anteroposterior stability constraint to predict MWC locomotion dynamics of the upper limbs during both startup and steady-state propulsion. Simulation results indicated the occurrence of MWC tipping when stability was not constrained, and that the constrained optimal control algorithm predicted different propulsion strategies. Hence, further proceedings of MWC locomotion simulation and optimal control investigations should take the anteroposterior stability into account to achieve more realistic simulations. Additionally, the implementation of the anteroposterior stability constrains unexpectedly resulted in a reduction of the computational time.