The complex nature of blood flow in the human arterial system is still gaining more attention, as it has become clear that cardiovascular diseases localize in regions of complex geometry and complex flow fields. In this article, we demonstrate that the lattice Boltzmann method can serve as a mesoscopic computational hemodynamic solver. We argue that it may have benefits over the traditional Navier-Stokes techniques. The accuracy of the method is tested by studying time-dependent systolic flow in a 3D straight rigid tube at typical hemodynamic Reynolds and Womersley numbers as an unsteady flow benchmark. Simulation results of steady and unsteady flow in a model of the human aortic bifurcation reconstructed from magnetic resonance angiography, are presented as a typical hemodynamic application.