Artificial soil is an important standard medium and reference material for soil ecotoxicity bioassays. Recent studies have documented the significant variability of their basic properties among different laboratories. Our study investigated (i) the variability of ten artificial soils from different laboratories by means of the fate, extractability and bioavailability of phenanthrene and lindane, and (ii) the relationships of these results to soil properties and ageing. Soils were spiked with (14)C-phenanthrene and (14)C-lindane, and the total residues, fractions extractable by hydroxypropyl-β-cyclodextrin, and the fractions of phenanthrene mineralizable by bacteria were determined after 1, 14, 28 and 56 days. Significant temporal changes in total residues and extractable and mineralizable fractions were observed for phenanthrene, resulting in large differences between soils after 56 days. Phenanthrene mineralization by indigenous peat microorganisms was suggested as the main driver of that, outweighing the effects of organic matter. Lindane total residues and extractability displayed much smaller changes over time and smaller differences between soils related to organic matter. Roughly estimated, the variability between the artificial soils was comparable to natural soils. The implications of such variability for the results of toxicity tests and risk assessment decisions should be identified. We also suggested that the sterilization of artificial soils might reduce unwanted variability.