The retention performances of a novel 80 wt% magnetite-silica composite HGMS matrix element and a traditional stainless steel wool HGMS matrix element were compared. Breakthrough and ultimate (strictly magnetic) retention experiments were carried out at different feed concentrations, flow velocities, and magnetic field intensities with an aqueous slurry containing iron oxide (Fe2O3) particles. Results from the ultimate retention experiments were also compared to those obtained from a potential flow model. Overall, when supported by silica in the form of the 80 wt% magnetite-silica composite, magnetite was about 30 to 40% as effective as stainless steel in retaining iron oxide, which was quite encouraging when considering that the magnetic saturation of magnetite is about five times smaller than that of stainless steel. Magnetite is also less expensive than stainless steel wool, even when encased in silica gel, which makes its use as an HGMS matrix element even more attractive for certain applications. The ultimate retentions predicted from the model for both matrix elements were of the same order of magnitude as the experimental results, an encouraging result: under predictions no greater than a factor of 7 for the stainless steel cylinder, and over predictions no greater than a factor of 3 for the magnetite spheres were obtained. The reasons for these discrepancies centered on the relatively wide size distributions of the Fe2O3 particles and both the magnetic matrix elements, which were not accounted for by the models. The potential flow model also could not account for the effects of laminar layers and separation zones on retention; nor could it account for the effects of the changing flow patterns caused by particle build-up on retention. Also, the models could not predict the dependence of the ultimate retention on the feed concentration observed experimentally, which suggested the existence of dynamic equilibrium at the external boundaries of the collection zones where the magnetic force was weaker.