Stable isotope ratios act as chemical tracers of animal diet, and are used to study food web dynamics. Because carbon stable isotope values are influenced by tissue lipid content, a number of extraction methods have been used to remove lipid bias, but, in some species and tissues, extractions also alter nitrogen isotope values. We have analyzed δ13C and δ15N in Atlantic bluefin tuna liver and white muscle, and whole Atlantic herring, fish tissues covering a wide range of lipid content (bulk C:N 3.1–12.5). In order to compare δ13C and δ15N values from traditional chloroform/methanol extractions with non-polar solvent alternatives, we analyzed samples following (1) no treatment, (2) lipid removal using chloroform/methanol (2:1), and (3) Soxhlet extractions using chloroform, diethyl ether or hexane. Chloroform/methanol and chloroform extractions produced the lowest C:N values and highest δ13C values. In bluefin tuna, chloroform and hexane extractions significantly altered liver δ15N, and all methods significantly altered δ15N values in white muscle. Whole Atlantic herring δ15N was not altered by any extraction method, while the 2:1 chloroform/methanol extraction most completely removed fish tissue lipid components. Our results indicate that δ15N effects are not limited to common chloroform/methanol extractions and suggest that chloroform/methanol is the most effective extraction for δ13C correction. Given evidence for δ15N alteration among all tested methods, mathematical correction approaches should be further explored as an alternative to lipid correction. Copyright © 2008 John Wiley & Sons, Ltd.