Thallium stable isotope data are used in this study, for the first time, to apportion Tl contamination in soils. In the late 1970s, a cement plant near Lengerich, Germany, emitted cement kiln dust (CKD) with high Tl contents, due to cocombustion of Tl-enriched pyrite roasting waste. Locally contaminated soil profiles were obtained down to 1 m depth and the samples are in accord with a binary mixing relationship in a diagram of Tl isotope compositions (expressed as ε(205)Tl, the deviation of the (205)Tl/(203)Tl ratio of a sample from the NIST SRM 997 Tl isotope standard in parts per 10(4)) versus 1/[Tl]. The inferred mixing endmembers are the geogenic background, as defined by isotopically light soils at depth (ε(205)Tl ≈ -4), and the Tl emissions, which produce Tl-enriched topsoils with ε(205)Tl as high as ±0. The latter interpretation is supported by analyses of the CKD, which is also characterized by ε(205)Tl ≈ ± 0, and the same ε(205)Tl value was found for a pyrite from the deposit that produced the cocombusted pyrite roasting waste. Additional measurements for samples from a locality in China, with outcrops of Tl sulfide mineralization and associated high natural Tl backgrounds, reveal significant isotope fractionation between soils (ε(205)Tl ≈ +0.4) and locally grown green cabbage (ε(205)Tl between -2.5 and -5.4). This demonstrates that biological isotope fractionation cannot explain the isotopically heavy Tl in the Lengerich topsoils and the latter are therefore clearly due to anthropogenic Tl emissions from cement processing. Our results thus establish that isotopic data can reinforce receptor modeling for the toxic trace metal Tl.