Purpose Traditionally, methods for sediment extractions are characterised using chemical analyses. However, in order to evaluate sediment extracts with regard to biological effects and, thus, bioaccessibility, extraction methods have to be compared to effect data obtained from experiments with in situ exposure scenarios, i.e., sediment contact tests. This study compares four extraction methods and sediment contact test data from a previous project with respect to predictive power in the fish embryo test with zebrafish (Danio rerio). Materials and methods A natural and an artificial sediment spiked with a mixture of six organic pollutants (2,4-dinitrophenol, diuron, fluoranthene, nonylphenol, parathion and pentachlorophenol) were extracted using (a) membrane dialysis extraction (MDE), (b) a Soxhlet procedure, (c) hydroxypropyl-β-cyclodextrin (HPCD) or (d) Tenax ® -TA. Whereas the former two are regarded being exhaustive with respect to non-covalently bound contaminants, the latter two are considered to predict bioaccessibility. Resulting extracts were tested in the fish embryo assay with D. rerio for embryotoxic and teratogenic potential. Results and discussion Mortalities caused by organic extracts from Soxhlet extraction and MDE were high. However, HPCD extracts turned out to be at least as effective as extracts obtained with these two methods. One possible reason might be short ageing time of the spiked sediments. Only Tenax ® -TA extracts gave results comparable to the sediment contact assay for natural sediment, but revealed low reproducibility. Significant differences between natural and artificial sediment were found for extracts obtained with techniques using native (i.e., nonfreeze-dried) sediments, i.e., HPCD and Tenax ® -TA. In contrast, MDE and Soxhlet extracts used freeze-dried sediment and did not differentiate between natural and artificial sediment. Therefore, freeze-drying has likely altered and equalised sediment properties that influence accessibility, such as composition of bacterial communities and organic matter quality. Conclusions Four extraction methods were successfully characterised with respect to their stringency and predictiveness for bioaccessibility. MDE was confirmed as an alternative to Soxhlet extraction. High mortalities induced by HPCD extracts underline the need to include ageing into consideration when assessing sediments. Although Tenax ® -TA may basically be used to predict bioaccessibility in the fish embryo test, the high variability observed warrants further investigation of the relation between effect and extractability. Apparently, freeze-drying can severely affect sediment properties, potentially eliminating individual properties of natural sediments.