Two methods are currently available for determining contaminant assimilation efficiencies (AE) from ingested material in benthic invertebrates. These methods were compared using the Great Lakes amphipod Diporeia spp. and [14C]benzo(a)pyrene (BaP) sorbed to Florissant sediment (< 63 µm). The first approach, the direct measurement method, uses total organic carbon as a tracer and yielded AE values ranging from 45.9~50.4%. The second approach, the dual-labeled method, uses 51Cr as a non-assimilated tracer and did not yield AE values for our data. The inability of the dual-labeled approach to estimate AEs was due, in part, to the selective feeding by Diporeia resulting in a failure of the non-assimilated tracer (51Cr) to track with the assimilated tracer ([14C]BaP). The failure of the dual-labeled approach was not a result of an uneven distribution of the labels among particle size classes, but more likely resulted from differential sorption of the two isotopically labeled materials to particles of differing composition. The [14C]BaP apparently sorbs to organic particles that are selectively ingested, while the 51Cr apparently sorbs to particles which are selectively excluded by Diporeia. The dual-labeled approach would be a viable and easier experimental approach for determining AE values if the characteristics that govern selective feeding can be determined.