The detection of toxins in shellfish through reliable methods is essential for human health preservation and prevention of economic losses in the aquaculture industry. Although no human intoxication has been unequivocally linked to gymnodimines or spirolides, these phycotoxins are highly toxic by intraperitoneal injection causing false positives in lipophilic toxin detection by the mouse bioassay. Based on the detection of molecular interactions by fluorescence polarization an inhibition assay was developed using fluorescent alpha-bungarotoxin and nicotinic acetylcholine receptor-enriched membranes of Torpedo marmorata to detect gymnodimine and 13-desmethyl C spirolide. Both toxins, classified into the cyclic imine group, inhibit the interaction of alpha-bungarotoxin with Torpedo nicotinic acetylcholine receptors in the nM range. In this study we analyze the matrix effect of four shellfish species on the fluorescence polarization assay. Mussels, clams, cockles and scallops were extracted with acetone and sequentially partitioned with n-hexane and chloroform. The interference of these shellfish extracts with the alpha-bungarotoxin fluorescence or its binding to the nicotinic acetylcholine receptor was lower than 11%. The average recovery rates of gymnodimine and 13-desmethyl C spirolide using these solvents were 90.6+/-7.8% and 89.6+/-3.2%, respectively with variations among species. The quantification range of this fluorescence polarization assay for gymnodimine and 13-desmethyl C spirolide in all tested species was 80-2000 microg kg(-1) and 85-700 microg kg(-1) of shellfish meat, respectively. This assay format can be used to detect gymnodimine and 13-desmethyl C spirolide in shellfish as a screening assay.