Intravoxel incoherent motion (IVIM) analysis typically assumes that the motion of blood caused by microcirculation mimics a random walk with several steps taken during the diffusion encoding (diffusive regime). Some studies have suggested use of the other extreme regime where no direction changes occur during diffusion encoding (ballistic regime). However, data available suggest that an intermediate regime is more likely. In this study, we explore the impact of assuming different IVIM regimes on modeling and parameter estimation. Results on healthy liver indicate that substantial bias may be introduced unless proper modeling is used.