To accurately assess potential nanotoxicity on the basis of cellular iron content, the precise separation of cells into subpopulations according to their magnetic nanoparticle loading is of crucial importance. In this study, we developed a microfluidic magnetophoresis device consisting of a trapezoidal channel containing five side outlet branches and a narrow rectangular channel with three outlet branches. This unique structure enabled the sequential separation of cells loaded with tiny amounts of iron oxide and cells heavily labeled with iron oxide, in a single device. As a proof of concept, we attempted the sequential separation of Raw 264.7 cells with a large heterogeneity in uptake capabilities (1–50 pg of iron per cell). Consequently, we were able to differentiate the bulk cell population into seven subpopulations according to their mean iron oxide loading. We also evaluated potential nanotoxicity effects using the production of excess reactive oxygen species (ROS) and the inhibition of proliferation on the separated subpopulations, and we found that 46.6% of cells loaded with iron above the threshold value (16.4 pg) had higher ROS levels than the control group. Cells loaded with more than 3.7 pg of iron exhibited transiently inhibited cell-cycle progression. In particular, cells loaded with more than 35.4 pg of iron exerted a significant effect on cell proliferation. The proposed system could be useful in the investigation of nanotoxicity effects of iron oxide nanoparticle-induced cells, based on their iron oxide nanoparticle loading.