Levels of cortisol in fish blood provide quantitative information on the degree of stress induced by a variety of stressors. It is also useful in describing the social status of individual fish within groups. The commercial production of radioimmunoassay (RIA) kits, such as the DPC(R) Coat-A-Count radioimmunoassay kit, has considerably reduced the effort required for cortisol measurement. These kits employ human plasma based cortisol standards which are not compatible for use with non mammalian species such as fish e.g. turbot, Scophthalmus maximus (Rafinesque), blood due to the interference effect of lipids and steroid binding proteins present in the plasma. In this study the DPC(R) kit was used following the removal of these lipids and steroid binding proteins from the plasma using an ethanol-hexane extraction. Excessive variability in the cortisol values obtained using this method deemed it unsatisfactory in overcoming the problem of incompatibility. A second modification of this technique that was tested involved the preparation of turbot specific standards for use in the preparation of modified standard curves. Using this method, an accuracy of 93.4% was achieved, as opposed to 79.6% using the kit human plasma based standards, and 47.1% using samples following lipid removal using an ethanol-hexane extraction. Based on analysis of accuracy, precision and reproducibility it is concluded that commercially available cortisol kits are suitable for use with turbot plasma, but a number of minor modifications are necessary. (C) 1999 Elsevier Science Inc. All rights reserved.