sField observations have shown that root residues maintain root-adhering soil for several months after harvest. The aim of this work was to compare post-harvest effect of Amaranthus hypochondriacus (amaranth), Phaseolus vulgaris (common bean) and Zea mays (maize) roots on root-adhering soil, aggregation and organic carbon content. The experimental site was located on a volcanic sandy soil (Typic Ustifluvent) in the Valley of Mexico. In 1999 and 2000, maize had the highest root mass (92 and 94 g m(-2)) and the highest root-adhering soil (9051 and 5876 g m(-2)) when a root-soil monolith of 0.20 m x 0.20 m x 0.30 m was excavated after harvest. In contrast, bean roots (2 and 5 g m(-2)) had only 347 and 23 g m(-2) of adhering soil per monolith in each year. Amaranth had intermediate values between maize and bean. Dry soil aggregate classes (< 0.25, 0.5, 1, 2, 5 and > 5 mm) were similarly distributed among the three species. The sum of the three soil macro-aggregates classes > 1 mm was 0.1 g g(-1) in both years. Neither water stability of the 2-5 mm aggregates (0.05-0.09 g g(-1)) nor soil organic C (SOC) in three aggregate classes (< 0.25, 1-2 and > 5 mm; mean 14.6 mg g(-1)) was affected by species (P < 0.05) in either year. Observations of thin sections (10 x and 40 x) revealed absence of macro-aggregates under maize. Soil compaction was attributed to high mass of maize roots in the sampled soil volume. Root systems sampled after harvest had the capacity to maintain a well structured soil mass, which was proportional to root mass. Root-adhering soil measured in the field could be used to select species promoting soil adhesion by roots. (c) 2005 Elsevier B.V. All rights reserved.