This paper is a side effect of preparing international publications on our long term research on soils’ susceptibility to wind erosion. For the paper to be internationally understandable we had to translate the texture classes from the Polish soil-agricultural maps (PTG1974), used as a basis to derive ten soil units investigated in the experiments, into the widely recognised USDA classification. We spotted that the PTG1974 classes of sandy soils, falling into USDA single SAND class, have large, reaching 1620% difference in deflation rates, 25% in the case of LOAMY SAND and SANDY LOAM class the difference was 300%. The differences of this magnitude within a single textural class imply that the USDA classes may be too general to be used in some domains of environmental modelling. This also implies that translating soil kinds (soil textural classes) in Polish soil-agricultural maps into the USDA textural classes is not rational and may lead to the loss of spatial variability of soil cover and the loss of credibility in modelling of environmental phenomena.