To test the impact of a range of long-term land use types on the partitioning of microbial residues among soil particles, samples from a Mollisol with plots under 100 years of continuous arable cropping, 30 years of simulated overgrazing to severely degraded bare soil, or 30 years of grassland restoration were investigated. The microbial residues, which were assessed among three particle-size fractions (<2, 2–20, and 20–250 μm) by amino sugar extraction, exhibited change with particle size and land use. Converting arable cropping to bare soil induced substantial depletion of amino sugars associated with the clay-size fraction, as a proportion of total carbon (C) and total soil mass, but not the silt- and sand-size fractions. Alternatively, switching arable soil to grassland increased amino sugar stocks in both the clay- and sand-size fractions. Analysis of the relative input of fungal and bacterial derived amino sugars indicated that fungal sources are the most dynamic with respect to land use change. These results highlight the selective vulnerability of microbial C pools in finer fractions under low plant C input and the selective recovery in specific fractions upon restoration, emphasizing the importance of the conversion of plant organic matter into mineral-associated microbial residues to promote stable soil organic C.