Over a period of 3 days saltwater mussels, Mytilus edulis, were fed a cyanobacteria, Microcystis aeruginosa, that contained a high concentration of microcystins. The mussels were killed on a periodic basis over the course of 2 months. Mussels were also collected at two sites were high levels of microcystins in tissues had been noted. A strategy based on the chemically unique nature of the C20 beta-amino acid, (2S,3S,8S,9S)-3-amino-9-methoxy-2,6,8-trimethyl-10-phenyldeca-4,6- dienoic acid (Adda), portion of the microcystins was used in conjunction with a protein phosphatase (PPase) assay to analyse for both covalently bound microcystins and free microcystins in the mussel tissues. The mussel PPase assay results were compared with the Lemieux oxidation gas chromatography-mass spectrometry (GCMS) analysis. Less than 0.1% of the total microcystin burden in the mussel tissue was found to be extractable with MeOH. Thus, direct evidence was provided for the existence of covalently bound microcystins in mussel tissues in vivo. The mussels rapidly cleared the covalently bound microcystins when transferred to untreated seawater. Within 4 days the total microcystin burden dropped from a high of 336.9 (+/- 45.8) micrograms/g wet tissue to 11.3 (+/- 2.6) micrograms/g. After 4 days postexposure until completion of the experiment the total levels remained below the detection limits of the GCMS method. The levels of free microcystins, extracted with MeOH and detected by the PPase assay, fell from 204 ng/g wet tissue to a residual 14 ng/g over a 53 day postexposure period. Presumably the bound microcystin present in the mussel tissue exists as a covalent complex with the PP-1 and PP-2A enzymes. We conclude that in any shellfish monitoring program it is the total tissue microcystin burden that needs to be considered.