Extrapolation of laboratory toxicity data to the field is hampered by differences in bioavailability, among other factors. Often, chemical availability of substances in the soil pore water is considered equivalent to bioavailability. However, for collembolans, which are not closely connected with the water phase of the soil, the situation may be different since other routes of exposure may also be of importance. In the present study, two species of Collembola, Folsomia candida Willem and Folsomia fimetaria L., were exposed to copper sulfate in different ways, for example, in aqueous solutions, in spiked soil, in soil from a copper-contaminated field site, and via the food supply (yeast). Reproduction of the collembolans was inhibited at elevated soil or yeast copper concentrations. The 10% effective concentration (EC10) values for freshly spiked soil were 50 [0.03-13,095] and 141 [20-1,1001] mg/kg for F. candida and F. fimetaria, and the corresponding values for yeast exposure were 2,900 [613-13,877] and 361 [0.1-972,864] mg/kg (95% confidence limits in square brackets). Soil from the contaminated field site had no effects on reproduction at copper concentrations up to 2,500 mg/kg. Internal copper concentrations were constant up to a soil copper concentration between 400 and 800 mg/kg in newly spiked soil and up to between 1,400 and 2,500 mg/kg in field soil. Copper accumulation in the animals from freshly spiked soil was higher than from yeast, and accumulation from both routes appeared to be additive. Spiked soil resulted in higher accumulation than soil from a contaminated field site. Addition of clean yeast to spiked soil resulted in a decrease in copper accumulation. The two collembolan species accumulated comparable concentrations from soil, whereas F. candida accumulated more copper from contaminated yeast placed on top of unpolluted soil than F. fimetaria. Male F. fimetaria accumulated more copper from contaminated soil than females. When exposed through yeast, the two sexes accumulated similar concentrations. A link between internal copper concentrations and effects on reproduction was difficult to establish because of large variations in both parameters and because effects seemed to begin at soil and yeast copper concentrations where internal concentrations were still regulated. Consequently, the establishment of a fixed internal threshold was uncertain.