An HPLC analytical method based on a methanol extraction, C-8 clean-up, derivatization using trifluoroacetic acid anhydride and reverse-phase HPLC was developed for the detection of 22,23-dihydroavermectin B (1a) in marine sediment. The limit of quantitation was determined as 0.93 ng 22,23-dihydroavermectin B (1a) (H (2) B (1a)) per g of sediment. This method was used to determine the concentrations of H (2) B (1a) in sediments taken from under a fish farm where ivermectin had been administered orally, at 0.05 mg/kg twice weekly, to control sea-lice infestation in 350,000 post-smelt Atlantic salmon. Prior to sampling, a total of 23 g of ivermectin had been administered during the current production cycle. The sediment was sampled using diver-collected cores at sample sites 25 m apart. The mean concentration of H (2) B (1a) detected in the six sediment cores taken directly under the cage block was 5.0 ng/g in the top 3 cm, 3.1 ng/g in the 3-6 cm fraction and 0.7 ng/g in the 6-9 cm fraction. Concentrations were under the limit of quantitation at depths below 9 cm in all cores taken from directly under the cage block. H (2) B (1a) was not detected in any samples taken further than 22.5 m from the cage block in the eastern, western and southern directions. To the north of the cage block, H (2) B (1a) was detected in samples taken 31 m from the cage block and the limit of distribution in this direction was not established. Despite the limitations of this study, it is difficult to reconcile the concentrations of H (2) B (1a) detected in this study with long term accumulation and persistence of ivermectin in sediments under fish farms. (C) 2000 Elsevier Science B.V. All rights reserved.