A liquid chromatography/tandem mass spectrometry (LC/MS/MS) method for the determination of fumonisin B1 (FB1), B2 (FB2) and B3 (FB3) in cornflakes is described. During method development, special attention was paid to the selection of a suitable internal standard (IS) in order to offer a good alternative for deuterated FB1. In this respect, the C12-sphinganine analogue (2S,3R)-2-aminododecane-1,3-diol was chosen because of its structural similarity to the fumonisin backbone and its chromatographic elution between the target analytes. For the extraction of the fumonisins from the cornflakes matrix, MeOH/H2O (adjusted to pH 4 with 0.1 M HCl; 70:30, v/v), ACN/MeOH/H(2)O (25:25:50, v/v/v) and acidified ACN/MeOH/H2O (25:25:50, v/v/v; pH 4) were evaluated. Preference was given to acidified MeOH/H2O (70:30, v/v) with mean recoveries (n=12) for FB1, FB2 and FB3 of, respectively, 84+/-10, 78+/-7 and 85+/-9%. Cleanup was performed using immunoaffinity columns (FumoniTest, VICAM). The chromatography was performed under isocratic conditions at a flow of 0.3 mL min-1 with a mobile phase consisting of ACN/H2O (60:40, v/v) containing 0.3% formic acid. The mass spectrometer was operated in the positive electrospray ionization (ESI+) mode using multiple reaction monitoring (MRM). An intralaboratory validation was conducted with fortified samples determining limits of detection (LOD), limits of quantification (LOQ), precision, trueness, specificity and measurement uncertainty. The LOD concentrations for FB1, FB2 and FB3 were 20, 7.5 and 12.5 microg/kg. The LOQs were 40 microg/kg for FB1, 15 microg/kg for FB2 and 25 microg/kg for FB3. The coefficients of variation (CVs) under repeatability conditions varied from 11 to 13% for FB1, from 9 to 14% for FB2 and from 7 to 10% for FB3. Under within-laboratory reproducibility conditions, the CVs ranged from 12 to 17% for FB1, from 9 to 16% for FB2 and from 7 to 13% for FB3. The percent bias for FB1 varied from -12 to -10%, while for FB2 and FB3 bias ranged, respectively, from -4 to -2% and from -12 to -5%. The expanded measurement uncertainties for FB1, FB2 and FB3 were, respectively, 19, 18 and 22%.