Soil sorption properties can influence the bioavailability of substances and consequently the toxicity for soil organisms. Current standardised laboratory testing for the exposure assessment of pesticides to soil organisms uses OECD artificial soil that does not reflect the high variation in chemical-physical soil properties found in natural agroecosystems. According to guideline OECD 222, earthworm reproduction tests with Eisenia fetida and the pesticide carbendazim were performed in four natural soils and OECD artificial soil. By using pF 1.6, which ensures a uniformity in actual soil water availability, the control reproduction performance of E. fetida in all natural soils was at the same level as OECD artificial soil. In a principle component analysis, the variation in toxicity between the tested soils was attributable to a combination of two soil properties, namely total organic carbon content (TOC) and pH. The largest difference of 4.9-fold was found between the typical agricultural Luvisol with 1.03% TOC and pH 6.2 (EC10: 0.17 (0.12–0.21) mg a.i. kg−1 sdw, EC50: 0.36 (0.31–0.40) mg a.i. kg−1 sdw) and OECD artificial soil with 4.11% TOC and pH 5.6 (EC10: 0.84 (0.72–0.92) mg a.i. kg−1 sdw, EC50: 1.07 (0.99–1.15) mg a.i. kg−1 sdw). The use of typical agricultural soils in standardised laboratory earthworm testing was successfully established with using the measure pF for soil moisture adjustment. It provides a more application-oriented approach and could serve as a new tool to refine the environmental risk assessment at lower tier testing or in an intermediate tier based approach.