The carry-over of dietary endosulfan to the fillet of farmed Atlantic salmon was studied. The uptake and elimination rate constants of the alpha and beta isoform of endosulfan were determined in seawater adapted Atlantic salmon (initial weigh 173+/-25 g) fed on endosulfan enriched diets (724 and 315 microg kg(-1) for alpha- and beta-endosulfan, respectively) for 92 days, followed by a 56 days depuration period with feeding on control diets (<0.3 microg kg(-1) endosulfan). The accumulation of the toxic metabolite endosulfan sulphate, which was not detected (<0.5 microg kg(-1)) in the experimental feeds, was also determined. Dietary beta-endosulfan was more persistent than alpha-endosulfan as demonstrated by a higher uptake (41+/-8% vs. 21+/-2%) and lower elimination (26+/-2 x 10(-3) day(-1) vs. 40+/-1 x 10(-3) day(-1)) rate constants, and a higher biomagnification factor (0.10+/-0.026 vs. 0.05+/-0.003, p<0.05). Based on the decrease in diastereometric factor over time, biotransformation was estimated to account for at least 50% of the endosulfan elimination. The formation of the metabolite endosulfan sulphate comprised a maximum 1.2% of the total accumulation of endosulfan. Using a simple one-compartmental model and the experimentally-determined rate constants, it was predicted that only dietary concentration higher than 7.5 mg kg(-1) would cause the estimated maximum residue level (MRL) in fillet of 0.47 mg kg(-1) to be exceeded. Model estimations also predict that current EU maximum permitted levels in fish feeds (0.005 mg kg(-1)) would give fillet concentrations that are under the detection limit (0.3 microg kg(-1)), which is in agreement with the general absence of endosulfan in monitored farmed Atlantic salmon.