Primary organo-mineral complexes are defined as organic matter (OM) bound to primary mineral particles, isolated after complete dispersion of soil. Organic matter present in < 2 µm particle-size fractions of soils has slow turnover times and it is assumed to be stabilized mainly by interaction with minerals. We aimed to quantify how much of the organic matter in < 2 µm particle-size fractions was free versus bound to minerals and to describe the nature of the association. Furthermore, we hypothesized that this bound organic matter was more resistant to biodegradation than free organic particles. We tested this by using a cultivation chronosequence on temperate silty soils and quantified free and clay-bound organic matter using density fractionation coupled with elemental analyses, as well as transmission electron microscopy (TEM) coupled with image analyses. Both methods showed that free organic matter was a minor fraction and that it was more depleted by cultivation than clay-bound organic matter. We deduced that clay-bound organic matter was more resistant to biodegradation. TEM showed that the distribution of organic matter in clay-sized fractions was patchy and that many of the so-called < 2 µm 'particles' were in fact nanometre- to micrometre-sized microaggregates in which OM was encrusted by minerals or coated minerals. We conclude that true primary organo-mineral complexes do not correspond to reality and must be regarded as conceptual entities. We suggest that the very small microaggregates, which were evidenced here, are major sites of OM stabilization, both by adsorption and by entrapment of organic matter.