Our results show that the measurement of cortisol levels from gill filaments is a reliable indicator of the general cortisolemia in the fish. The correlation between the cortisol levels in plasma and gill is good, in animals exposed of not to stress. For zebrafish, the plasma-gill cortisol correlation was not as good as in trout, probably reflecting the lower number of fish used and also the difficulties associated with the blood collection procedure. Regarding the latter, the tiny volume of blood sample taken could result in a higher chance of contamination of the sample with water of other body fluids. For the reason, we tested two different ways of normalizing plasma cortisol data in zebrafish: by plasma volume (as in the trout) or by plasma protein content. The latter showed better performance and was therefore utilized for graphs and statistical analyses. Similarly, in the case of the gill, we tested three different ways to normalize gill cortisol data in both trout and zebrafish: by gill wet mass, by gill protein content or by gill hemoglobin content. Hemoglobin was used ans an indicator of the amount of blood within the gill, since it is known that stress can induce structural alterations in the gill (Wendelaar Bonga, 1997) and that could lead to differences in the gill blood content between control and stressed fish (and that would have been a jeopardizing factor when trying to assess stress from gill cortisol). Although our results showed that the levels of gill hemoglobin were not altered by stress, the normalization by gill protein content rendered better results and therefore, was used for statistical analyses and for the graphs.