Within the scope of a phytoremediation project, the toxicity of 2,4,6-trinitrotoluene (TNT) contaminated soil (and its toxic metabolites) on earthworms was assessed. In addition to the standard acute and reproduction tests (ISO 11268), an avoidance response test was applied. The test methods covered all important ecological relevant endpoints (acute, chronic, behavioral). At a concentration of 1142 mg/kg, TNT caused significant toxic effects in all test methods, but at lower test concentrations no significant acute or reproduction toxic effects could be observed. The avoidance response test, however, showed significant repellent effects at a concentration of 29 mg/kg TNT and therefore proved to be more sensitive than the other tests in this case. Results of the earthworm tests compared well with results of an ecotoxicological biotest battery. Thus, earthworm toxicity tests are useful tools for terrestrial risk assessment but require a hierarchy of test designs that differ in effect levels (behavior, sublethal, lethal). Whereas higher concentrations of a pollutant can easily be assessed with the acute test (which requires lethal concentrations to show an effect), contaminated soils with lower (sublethal) pollutant concentrations require more sensitive test methods such as reproduction or behavioral tests in their risk assessment.