Sandy soils (Typic Ustifluvent) in the Valley of Mexico are prone to erosion due to their structural instability. The objective of the work was to compare the capacity of roots of three crop species (amaranth, maize and sunflower) to maintain a sandy soil adhered to roots. Soil was sampled two and six months after crop harvest (February and June 2004) using a standard technique of soil extraction (20 x 20 x 30 cm). The root mass (RM) and root-adhering soil (RAS) were determined for each sample. Dry and water aggregate-stability tests were performed. Microstructure in the root-zone was also evaluated in February 2004, by image analysis on thin sections prepared from referenced root-soil samples. Two positions (centre of root-stalk axis, and a lateral position separated 11 cm from stalk), and four soil depths (0-7, 7-14, 14-21 and 21-28 cm) were sampled. Thin sections were obtained according to Bullock et al. (1985). Number of roots cm-2 (diameter > 2 mm), macro-aggregates cm-2 (diameter > 2 mm) and porosity were evaluated on thin sections using a video camera coupled to an Image Pro® system. Data were analyzed using the SAS program. We found that the three crops had similar values of root-adhering soil (5.5 kg m-2) in the first sampling date; however, the RAS/RB ratio was higher in maize (239) as compared to amaranth and sunflower (P<0.05). A general decay of root-adhering soil was observed between the first and second sampling date, but the decay was higher for amaranth (0.4 kg m-2 in the second date against 5.8 kg m-2 in the first one). The lower C:N ratio previously observed in amaranth might explains this result. Aggregate water-stability was also higher for maize soil samples. Number of roots cm-2 was affected by species, and by species x position interaction. Sunflower had the highest mean value in both sampling positions (0.32 cm-2 for root-centered and 0.19 cm-2 for lateral position). Means of root size were 4.1, 3.5 and 3.2 mm for sunflower, maize and amaranth (P<0.05). Position and depth factors did not affect these variables. A higher number of macro-aggregates cm-2 was found in surface (0-7 cm; P<0.05), as compared to sub-surface depths; non statistical difference by species was found. At 0-7 cm depth was found 1.35, against 0.66, 0.60 y 0.56 macro-aggregates cm-2 in the sub-surfaces strata. Crop species did not affect soil porosity measured on thin sections. Porosity in samples from the three species diminished with soil depth in both root-centered and lateral samples. Results of this work suggest that the structure of sandy soil can beneficiates from buried roots which decomposition (mainly in maize and sunflower) is very slow, due to low microorganism activity, coincident with low temperatures and low soil water content during the post-harvest period (January-May). The main conclusions are: a) roots of crop species presented differences in their capacity to adhere soil; roots of maize appeared with the highest RAS/RB value, and consequently as the more efficient adhering-soil roots; b) root adhering soil decayed with time sampling; maize and sunflower had more recalcitrant roots as compared to amaranth, and c) increased values of macro-aggregate density and soil porosity were found in surface soil (0-7 cm) as compared to sub-surface strata; both variables were not affected by crop species.