Quantitative analysis of Scanning Electron Microscopy (SEM) images of soil fabric is a common technique for relating soil microstructure to macroscopic characteristics. The magnification at which a soil sample is investigated affects the accuracy and validity of interpretations of acquired data because most soils are hierarchical in structure. Patterns that are evident at one magnification and frame size may be absent in other magnification and frame size ranges. Thus, there is a magnification effect in the visualization of soil fabric and estimation of soil characteristics that are known to depend on soil structural arrangements. In this paper, an expansive soil fabric was studied using progressive magnification and enlargement of frame sizes to define size proportion values ranging from about 0.01 to 1.0 and using binary indexing (1 for aggregate and 0 for pore space) to develop a matrix. The matrix values were then combined with fractal analysis to relate magnification/frame size to planar porosity and fractal dimensions of soils. The results for an expansive soil from Nanyang, P.R. China show that, when the size proportion is less then 0.3, there is excessive scatter in estimated porosities. Aggregate orientation has less sensitivity to size proportion, and the box-counting fractal dimension increases sharply as the size proportion decreases to levels below 0.1.