Recent studies raised increasing concern about the reliability of computer models in reproducing in vivo hemodynamics in cerebral aneurysms. Boundary condition problem is among the most frequently addressed issues since three-dimensional (3-D) modeling is usually restricted to local arterial segments. The present study focused on aneurysms on the anterior communicating artery (ACoA) which represent a large subgroup of detected cerebral aneurysms and, in particular, have a relatively high risk of rupture compared to aneurysms located in other regions. The sensitivity of blood flows in three ACoA aneurysms to boundary conditions was investigated using 3-D hemodynamic models. The boundary conditions of the 3-D models were predicted by a one-dimensional (1-D) model of the cerebral arterial network. The parameters of the 1-D model were assigned based respectively on population-averaged data and patient-specific data derived from medical images, yielding a population-generic model and a patient-specific model. In addition, particle image velocimetry (PIV) experiments were performed to validate the code used to simulate intra-aneurysmal blood flows. Obtained results showed that switching the boundary conditions of the aneurysm models from population-generic ones to patient-specific ones led to pronounced changes in simulated intra-aneurysmal flow patterns in terms of vortex structure, impingement region and the magnitude and spatial distribution of wall shear stress and oscillatory shear index. In particular, the way and the degree in which hemodynamic quantities are influenced by boundary conditions exhibited pronounced inter-patient variability. In summary, our study underlines the importance of patient-specific treatment of boundary conditions in model studies focusing on ACoA aneurysms.