Horizontal and vertical root mapping was used to relate rooting to a structured soil environment. Selected horizontal planes and associated vertical walls were exposed, and root locations were marked on polyethylene sheets. Separate sheets were used to indicate pores and cracks. Image analysis of the marked sheets indicated location of each root and root number per unit area (N). Number of roots growing in pores and cracks was determined by alignment of the marked sheets. Sixteen undisturbed cores (118 cm3) were taken in a grid pattern from each Ap horizon and adjacent subsoil for determination of root length density (D). Root clustering was associated with high coefficients of variation (ranging from 27 to 102%) for D. D/N (number−1) ratios were 16 (much greater than the theoretical value of 2 for a random root system). Both root branching and location of plant rows caused root clustering. Most of the root mappings were considered clustered by distributional statistics methods. Only a small percentage of roots were growing in preexisting pores and cracks, but many roots were located on ped faces.