Measures (e.g. δ15N, δ13C, %C, %N and C:N) derived from animal tissues are commonly used to estimate diets and trophic interactions. Since tissue samples are often exposed to air or kept chilled in ice over a short-term during sample preparation, they may degrade. Herein, we hypothesize that tissue decomposition will cause changes in these measures. In this study, we kept marine fish, crustacean and mollusc tissues in air or ice over 120 h (5 days). We found that tissue decomposition in air enriched δ15N (range 0.6‰ to 1.3‰) and δ13C (0.2‰ to 0.4‰), decreased %N (0.47 to 3.43 percentage points from staring values of ~13%) and %C (4.53 to 8.29 percentage points from starting values of ~43%), and subsequently increased C:N ratio (0.14 to 0.75). In air, while such changes to δ13C were relatively minor and therefore likely tolerable, changes in δ15N, %N, %C and C:N ratio should be interpreted with caution. Ice effectively reduced the extent to which decomposition enriched δ15N (≤ 0.4‰) and δ13C (≤ 0.2‰), and eliminated decomposition in C:N ratio, %N and %C. In our second experiment, for fish tissues in either air or ice over 120 h, we observed no effects of decomposition on relationships between lipid content, C:N ratio, and Δδ13C (change in δ13C after lipid removal), which are employed to correct δ13C for samples containing lipid. We also confirmed that lipid in tissues caused large errors when estimating δ13C (mean ± standard error = -1.8‰ ± 0.1‰, range -0.6‰ to -3.8‰), and showed both lipid extraction and mathematical correction performed equally well to correct for lipids when estimating δ13C. We, therefore, recommend that specimens of marine animals should be kept in ice during sample preparation for a short-term, as it is an effective means for minimizing changes of the stable isotope measures in their tissue.