The accuracy, repeatability, and reproducibility characteristics of a liquid chromatographic method for the determination of ochratoxin A (OTA) in white wine, red wine, and beer were established in a collaborative study involving 18 laboratories in 10 countries. Blind duplicates of blank, spiked, and naturally contaminated materials at levels ranging from < or =0.01 to 3.00 ng/mL were analyzed. Wine and beer samples were diluted with a solution containing polyethylene glycol and sodium hydrogen carbonate, and the diluted samples were filtered and cleaned up on an immunoaffinity column. OTA was eluted with methanol and quantified by reversed-phase liquid chromatography with fluorometric detection. Average recoveries from white wine, red wine, and beer ranged from 88.2 to 105.4% (at spiking levels ranging from 0.1 to 2.0 ng/mL), from 84.3 to 93.1% (at spiking levels ranging from 0.2 to 3.0 ng/mL), and from 87.0 to 95.0% (at spiking levels ranging from 0.2 to 1.5 ng/mL), respectively. Relative standard deviations for within-laboratory repeatability (RSDr) ranged from 6.6 to 10.8% for white wine, from 6.5 to 10.8% for red wine, and from 4.7 to 16.5% for beer. Relative standard deviations for between-laboratories reproducibility (RSDR) ranged from 13.1 to 15.9% for white wine, from 11.9 to 13.6% for red wine, and from 15.2 to 26.1% for beer. HORRAT values were < or =0.4 for the 3 matrixes.