Life cycle impact assessment aims to translate the amounts of substance emitted during the life cycle of a product into a potential impact on the environment, which includes terrestrial ecosystems. This work suggests some possible improvements in assessing the toxicity of metals on soil ecosystems in life cycle assessment (LCA). The current available data on soil ecotoxicity allow one to calculate the chronic terrestrial HC50(EC50) (hazardous concentration affecting 50% of the species at their EC50 level, i.e. the level where 50% of the individuals of the species are affected) of nine metals and metalloids (As(III) or (V), Be(II), Cr(III) or (VI), Sb(III) or (V), Pb(II), Cu(II), Zn(II) and Ni(II)). Contrarily to what is generally advised in LCIA, the terrestrial HC50 of metals shall not be extrapolated from the aquatic HC50, using the Equilibrium Partitioning method since the partition coefficient (Kd) of metals is highly variable. The experimental ecotoxicology generally uses metallic salts to contaminate artificial soils but the comparison of the EC50 or NOEC obtained for the same metal with different salts reveals that the kind of salt used insignificantly influences these values. In contrast, depending on the metallic fraction of concern, the EC50 may vary, as for cadmium: the EC50 of Folsotnia candida, expressed as free Cd in pore water is almost 2.5 orders of magnitude lower than that expressed as total metal. A similar result is obtained with Eisenia fetida, confirming the importance of metals speciation in assessing their impact on soils. By ranking the metals according to the difference between their terrestrial and aquatic HC50 values, two groups are distinguished, which match the hard soft acids and bases (HSAB) concept. This allows to estimate their affinity for soil components and potential toxicity according to their chemical characteristics. (C) 2007 Elsevier Ltd. All rights reserved.