The maturation of magnetic cell separation technology places increasing demands on magnetic cell separation performance. While a number of factors can cause sub-optimal performance, one of the major challenges can be non-specific binding of magnetic nano- or microparticles to non-targeted cells. Depending on the type of separation, this non-specific binding can have a negative effect on the final purity, the recovery of the targeted cells, or both. In this work, we quantitatively demonstrate that non-specific binding of magnetic nanoparticles can impart a magnetization to cells such that these cells can be retained in a separation column and thus negatively impact the purity of the final product and the recovery of the desired cells. Through experimental data and theoretical arguments, we demonstrate that the number of MACS magnetic particles needed to impart a magnetization that is sufficient to cause non-targeted cells to be retained in the column to be on the order of 500–1,000 nanoparticles. This number of non-specifically bound particles was demonstrated experimentally with an instrument, cell tracking velocimeter, CTV, and it is demonstrated that the sensitivity of the CTV instrument for Fe atoms contained in magnetic nanoparticles on the order of 1 × 10−15 g/mL of Fe. Biotechnol. Bioeng. 2010;105: 1078–1093. © 2009 Wiley Periodicals, Inc.