Five bioassays (inhibition of lettuce germination and growth, earthworm mortality, inhibition of springtail population growth, avoidance by springtails) were compared, using four coke factory soils contaminated by PAHs and trace elements, before and after biotreatment. For each bioassay, several endpoints were combined in an 'ecoscore', a measure of test sensitivity. Ecoscores pooled over all tested bioassays revealed that most organisms were highly sensitive to the concentration of 3-ring PAHs. When four soils were combined, behavioural tests using the springtail Folsomia candida showed higher ecoscores, i.e. they were most sensitive to soil contamination. However, despite overall higher sensitivity of behavioural tests, which could be used for cheap and rapid assessment of soil toxicity, especially at low levels of contamination, some test endpoints were more sensitive than others, and this may differ from a soil to another, pointing to the need for a battery of bioassays when more itemized results are expected.