Fresh plant residues are often identified with the light organic matter obtained by fractionation in a dense liquid, but the extent to which residues can be extracted efficiently by densimetric methods has not been widely studied. This paper presents the results of two experiments in which 14C-labelled straw was mixed with two soils of differing texture, and the mixture was subjected to densimetric fractionation. In the first experiment, soil was mixed with bulk milled straw, and in the second, with straw of different sizes (> 200 μm, 200-50 μm, 50-20 μm and < 20 μm). The recovery of straw in the light fraction was low, and decreased with decreasing straw size, reaching a minimum (less than 20 %) for straw < 20 μm. Except for straw < 20 μm, the recovery in the light fraction was lower in soil richest in silt and clay. In the absence of soil (blanks), the recovery of fine straw in the light fraction was lower than in the presence of soil, suggesting that the recovery of fine straw in the light fraction is partly due to its association with light coarse debris. Extractable and polytungstate-soluble fractions accounted for a small proportion of the 14C-activity. These results suggest that densimetric methods are not efficient for recovering fresh plant residues, except in the case of large residues in coarse-textured soils.