In a recent paper by Polyakov and Nearing (2003) it was shown experimentally that the sediment transport capacity in a rill is not unique for a given soil type, slope, and flow rate. Indeed, they found that the transport capacity was dependent on whether sediment transport in the rill was occurring under net erosion or net deposition conditions. They concluded that this nonuniqueness in transport capacity is a discrepancy that needs addressing in soil erosion models. Here we postulate that this behavior occurs as a result of defining transport capacity as an model input to distinguish between net erosion and net deposition regimes, instead of determining it as an outcome between the separate but continuous rate processes of deposition and entrainment such as is the case for the multisize class erosion model of Hairsine and Rose (1992a, 1992b). This model is used to reinterpret and reproduce the results of Polyakov and Nearing (2003). The analysis shows that the transport capacity cannot be unique for a soil composed of a range of size classes and that uniqueness only occurs for the exceptional case of single size class soil. Consequently, when used as a model input, the transport capacity concept is deficient in modeling sediment transport of real soils across different flow conditions.