Isolated perfused preparations permit the use of physiological routes of exposure for the study of mechanisms of toxicity under conditions where the architecture remains intact and the potential influences of other organs are eliminated. The perfusion of gills in the isolated head preparation has been used extensively in aquatic toxicology. Quantitative comparisons between in vivo systems and this preparation are required for the extrapolation of results. This paper compares data on uptake and elimination rates obtained in vivo on rainbow trout with those provided by the perfusion of gills in the isolated head preparation. The use of compartmental models allowed calculations of equivalent rates, based on relative changes in the total amount of chemical. Similar absorption rates of nitrites, 3.0 +/- 0.7 v. 4.4 +/- 1.1 for in vivo and in vitro systems, respectively, were observed. The in vivo rate of nitrite elimination was 0.057 +/- 0.024; the concentration changes observed in the in vitro experiments on nitrite elimination were too low to estimate accurate rates, but data suggest, at least, a similar order of magnitude. For copper, the in vitro elimination rate, 0.33 +/- 0.04, is two orders of magnitude higher than in vivo, 0.0035 +/- 0.0005. This difference could be explained by the binding of copper to proteins and cells, suggesting this binding as the limiting factor for the elimination of copper. The mass balance method proposed in this paper was useful to compare gill perfusion data and in vivo compartmental assessments under quantitative estimations.