The study of pedodiversity and soil richness depends on the notion of soils as discrete entities. Soil classifications are often criticized in this regard because they depend in part on arbitrary or subjective criteria. In this study soils were categorized on the basis of the presence or absence of six lithological and morphological characteristics. Richness vs. area relationships, and the general pattern of soil variability and diversity, were then compared to analyses of pedodiversity based on Soil Taxonomy. The study area consists of sixteen 0.13-ha plots on forested sideslopes of the Ouachita Mountains, Arkansas, with a minimum of 20 classified soil pits per plot. An ad hoc classification was developed, from the standpoint of soil geomorphology and studies of the coevolution of soils and landscapes, and based on the regional environmental framework. Soils were classified based on (1) underlying geology (shale, sandstone bedrock, or transported sandstone rock fragments), and on the presence or absence of (2) texture contrast subsoils, (3) eluvial horizons, (4) surface and/or subsurface stone lines or zones, (5) lithological contrasts between soil and underlying geology, and (6) redoximorphic features. The soil geomorphic classification (SGC) yielded 40 different soil types (out of 288 possible different combinations of the criteria), compared to 19 different series or taxadjuncts identified by standard soil classification. However, 21 of the SGC soil types had only one or two representatives. Individual plots contained five to 11 different SGC soil types with extensive local variability. A standard power-function relationship between soil richness ( S ) and area or number of samples ( A ) provided the best fit for most plots ( S = cA b ). The exponent b was slightly higher than for the taxonomy-based analysis, but in general the analyses lead to similar conclusions with respect to the relationship between richness and area, and the relative importance of local, within-plot versus regional, between-plot variability. Results support the view that soils can be viewed and treated as discrete entities, that richness assessments are not necessarily extremely sensitive to the classification used, and that highly localized variability may be critical to pedodiversity. The suggested criteria for identifying discrete soil types are given, based on qualitative morphological differences and state factor relations, contiguity, and connectivity.