Five soils from Austria that have high contents of arsenic (As)due to anthropogenic and/or geogenic sources were separated intosand, silt, and clay fractions. The distribution of As among theparticle-size fractions was clay >> silt > sand. Theparticle-size fractions were analysed according to an 8-stepsequential extraction procedure (SEP) designed to assess theprimary reservoirs and extractability of As. Steps 1 and 2,(i.e. NH4NO3 and (NH4)2SO4) extractedthe least As, while considerably greater amounts were extractedin steps 3, 4 and 5 (i.e. NH4H2PO4,NH2OH-HCl, and NH4F). The greatest amounts of As wereextracted in steps 6 and 7, both NH4-oxalate extractantstargeting the amorphous and crystalline oxide fractions. Theresidual fraction (step 8) was typically low. The extraction ofthe clays contained generally more As in steps 3 to 8, while Asrelease in the first two steps was typically highest in the sandand silt fractions. These findings are in accord with factoranalysis indicating that the extractants of the SEP areseparated into weak (steps 1 and 2) and strong (steps 3 to 8)extractants. SEP data along with microprobe analyses (X-raymicroanalyses) indicate that most As is associated with Fe andis therefore not readily mobile. Anyhow, As could be releasedupon changes in redox potential or pH. The greaterextractability of As from sand fractions using weak extractantsmay explain the higher toxicity of As in coarse-textured soils,as found in previous studies.