The chemical composition of soil from the Glasgow (UK) urban area was used to identify the controls on the availability of potentially harmful elements (PHEs) in soil to humans. Total and bioaccessible concentrations of arsenic (As), chromium (Cr) and lead (Pb) in 27 soil samples, collected from different land uses, were coupled to information on their solid-phase partitioning derived from sequential extraction data. The total element concentrations in the soils were in the range <0.1–135mgkg–1 for As; 65–3680mgkg–1 for Cr and 126–2160mgkg–1 for Pb, with bioaccessible concentrations averaging 27, 5 and 27% of the total values, respectively. Land use does not appear to be a predictor of contamination; however, the history of the contamination is critically important. The Chemometric Identification of Substrates and Element Distribution (CISED) sequential chemical extraction and associated self-modelling mixture resolution analysis identified three sample groupings and 16 geochemically distinct phases (substrates). These were related to iron (n=3), aluminium–silicon (Al–Si; n=2), calcium (n=3), phosphorus (n=1), magnesium (Mg; n=3), manganese (n=1) and easily extractable (n=3), which was predominantly made up of sodium and sulphur. As, Cr and Pb were respectively found in 9, 10 and 12 of the identified phases, with bioaccessible As predominantly associated with easily extractable phases, bioaccessible Cr with the Mg-dominated phases and bioaccessible Pb with both the Mg-dominated and Al–Si phases. Using a combination of the Unified Barge Method to measure the bioaccessibility of PHEs and CISED to identify the geochemical sources has allowed a much better understanding of the complexity of PHE mobility in the Glasgow urban environment. This approach can be applied to other urban environments and cases of soil contamination, and made part of land-use planning.