Measurements of depth-dependent cortical activity provide insights on directional activity between brain areas. While previous studies demonstrated the feasibility of human depth-dependent fMRI, the stability and reliability of depth-dependent results are less studied. In this work, we investigate sources of inconsistencies in depth-dependent activity profiles. We find that depth-dependent activity profiles are highly reproducible across different scanning sessions. They are, however, quite variable within cortical areas across different cross sections along the cortical ribbon. Only when depth-dependent profiles are considered with respect to their location along the cortical ribbon, task-driven modulations of input-output activity become consistent across participants.