In this work, the inorganic chemical profiles of soil samples collected at different sites in the Salentum peninsula (Italy, Apulia region) are discussed. The samples were re-suspended in the laboratory, for PM10 sampling, using a ventilated wooden chamber and were then chemically analysed measuring the abundances of 17 elements. Different land use categories of soils (olive grove, arable land, vineyards, sand, and urban dust) were included in the 50 samples analysed: 45 collected in background areas and five collected in the urban area of Lecce. The objectives were to compare the chemical profiles of raw soil and re-suspended PM10 for different crustal sources and to estimate the potential improvements in the calculation of the enrichment factors of atmospheric PM10. The variability of elemental abundances in samples of the same category of soil collected in different zones was of the same order of magnitude as the differences observed between the various categories of soil. This allows the calculation of a weighted average composition of soil and urban dust and the corresponding weighted average composition of re-suspended PM10. In re-suspended PM10 from average background soil, all of the elements except Ca, Na, K and V have larger abundances with respect to raw soil. In urban dust, this is limited to Ca, V and Mg. The crustal enrichment factors (EFs) of atmospheric PM10 were evaluated by considering different reference elements and different reference tables. Results indicated that it is possible to apply a two-threshold (S1 and S2) scheme for the interpretation of EF, with thresholds derived from uncertainty in soil categories and from the choice of the reference element. A specific element is likely of crustal origin if EF < S1 and likely of anthropogenic origin if EF > S2. Between the two thresholds, the element can be considered of mixed origin. The thresholds vary according to the geological composition used in the evaluation of EF. If the average composition of local re-suspended soils is used, the thresholds are S1 = 2 and S2 = 4. If raw soil is used, the thresholds become S1 = 5 and S2 = 10. If the average upper-crust composition from literature data is used, the thresholds further increase to S1 = 10 and S2 = 20.