Mineral magnetic properties of soils and parent materials have been interpreted in terms of paleoclimate and rates of soil formation but it is important to understand which minerals contribute to the mineral magnetic signal. Citrate-bicarbonate-dithionite (CBD) treatment has been used to determine the amounts of fine-grained, often pedogenic, ferrimagnetic minerals relative to coarse-grained, often inherited, magnetic minerals. However, questions have been raised about the effect of particle size on the efficacy of CBD in dissolving magnetite and maghemite grains. In this paper we use magnetic susceptibility and its frequency dependence, and the low-temperature behavior of a saturation isothermal remanent magnetization, to track the dissolution of carefully sized magnetite grains. We found that the standard CBD procedure dissolves fine magnetite particles (ca. < 1 μm) but leaves larger particles (ca. > 1 μm) essentially intact. Thin oxidized coatings, presumably maghemite, are also dissolved by the CBD procedure. These results support previous interpretations that the CBD procedure can be used to distinguish between pedogenic and lithogenic magnetic grains, assuming that most pedogenic magnetic grains are sufficiently small (ca. < 1 μm) and most lithogenic magnetic grains are sufficiently large (ca. > 1 μm). These results also show that the standard procedure is too harsh to differentiate between 1 μm grains of magnetite and maghemite. A modified CBD extraction that uses half as much dithionite reduces the magnetic susceptibility of 1 μm magnetite grains by only 10%. This method may be useful in distinguishing between magnetite and maghemite grains in this size range.