In marine food web studies, stable isotopes of nitrogen (delta N-15) and carbon (delta C-13) are widely used to estimate organisms' trophic levels (TL) and carbon sources, respectively. For smaller organisms, whole specimens are commonly analyzed. However, this "bulk method" involves several pitfalls since different tissues may fractionate stable isotopes differently. We compared the delta N-15 and delta C-13 values of exoskeleton versus soft tissue, in relation to whole specimens, of three common Arctic amphipods in Svalbard waters: the benthic Anonyx nugax, the sympagic (ice-associated) Gammarus wilkitzkii and the pelagic Themisto libellula. The delta N-15 values of the exo-skeletons were significantly lower than those of the soft tissues for A. nugax (10.5 +/- 0.7 parts per thousand vs. 15.7 +/- 0.7 parts per thousand), G. wilkitzkii (3.3 +/- 0.3 parts per thousand vs. 8.3 +/- 0.4 parts per thousand) and T. libellula (8.6 +/- 0.3 parts per thousand vs. 10.8 +/- 0.3 parts per thousand). The differences in delta C-13 values between exoskeletons and soft tissues were insignificant, except for A. nugax (-21.2 +/- 0.2 parts per thousand vs. -20.3 +/- 0.2 parts per thousand, respectively). The delta N-15 values of whole organisms were between those of the exoskeletons and the soft tissues, being similarly enriched in N-15 as the exoskeletons (except G. wilkitzkii) and depleted in N-15 by 1.2-3.7 parts per thousand compared to the soft tissues. The delta N-15-derived TLs of the soft tissues agreed best with the known feeding preferences of the three amphipods, which suggest a potential underestimation of 0.5-1.0 TL when stable isotope analyses are performed on whole crustaceans with thick exoskeletons. The insignificant or small differences in delta C-13 values among exoskeletons, soft tissues and whole specimens, however, suggest low probability for misinterpretations of crustaceans' primary carbon source in marine ecosystems with distinctly different delta C-13-carbon sources.