Laboratory experiments were conducted using intact collembolan communities, exposed to Madit D-(R) a phenylurea herbicide (active ingredient isoproturon). Effects were investigated using two distinct humus types, an acid Dysmoder and a neutral Eumull. Within two weeks, no effect of the herbicide was displayed by the Eumull population, while the Dysmoder population was stimulated. When animals were able to escape from the herbicide through a perforated wall separating two compartments filled with natural soil, the behavior of collembolan communities exhibited interactive (non-additive) effects of humus type and herbicide application. The combination of an acid soil (supposedly providing greater tolerance to organic pollutants) with a neutral soil, increased biodiversity of Collembola, but caused the disappearance of some acido-sensitive species, pointing to complex relationships between pesticides, soils and soil organisms. Parallel experiments with single species demonstrated that at the recommended dose Madit D-(R) may cause avoidance effects, but no toxicity.