In this study, the Cd isotopic composition of various geological reference materials and anthropogenic samples was investigated. The measurements were made by multicollector ICP-MS and instrumental mass fractionation was controlled using a "sample-standard bracketing" technique. Cadmium isotopic data are reported relative to an internal Cd solution (cd Spex) and expressed as the Cd-114/Cd-110 delta value. Two other Cd solutions (Prolabo and JMC) were analysed and yielded the same 0 parts per thousand delta value. A fractionated Cd metal sample (Monster Cd) was used as a secondary reference material for Cd isotopic measurements and we obtained a Cd-114/Cd-110 delta value of 4.48 parts per thousand relative to Cd Spex solution. As opposed to multi-stage Cd purification previously published in the literature, a new one step anionic exchange purification using dilute HCl for the analysis of Cd isotopes in geological samples was developed. This method enabled a high recovery (> 95%) and effective separation of the sample matrix to be achieved. The long-term external reproducibility was evaluated at 0.12 parts per thousand (2 standard deviations) for the Cd-114/Cd-110 ratio, based on reference solutions and replicated measurements of samples over one year. The variation of Cd isotopic composition of natural terrestrial samples is restricted to a small range of 0.4 parts per thousand, which is similar to previously reported results. In contrast, large variations of Cd isotopic composition were found for anthropogenic samples with values as low as -0.64 parts per thousand for a dust sample issued from a lead smelter and values as high as +0.50 parts per thousand for NIST SRM 2711 (metal-rich soil). These variations are 10 times larger than thereproducibility and suggest that Cd isotopes can be useful as tracers of anthropogenic sources of Cd in the environment.