The sources of variation In the analysis of trichothecenes in cereals by gas chromatography mass spectrometry (GC-MS) were studied and Identified. Ways to decrease some of the variations identified are presented and discussed. The method Is validated for deoxynivalenol, nivalenol, 3-acetyldeoxynivalenol, fusarenon-X, diacetoxyscirpenol, HT-2 toxin and T-2 toxin. The general variability in the performance of the GC-MS apparatus itself, as well as derivatisation, matrix effect, injection and quantification were identified to be the sources of variation. The use of internal standards was studied in order to decrease the variation due to problems with derivatisation and the overall sample preparation. Two derivatisation reagents were compared, one of which was found to be more effective for decreasing this variation. The calibrants prepared in the cereal extract reduced the variation due to the matrix effect and thus improved the quantification. "Constant Standards" were introduced in order to detect and decrease the variation caused by injection and the apparatus itself. The validation study proved that this analytical method for trichothecenes was adequately reliable and sensitive. The relative standard deviation varied between 3-9% and the recovery between 46-90% for the different trichothecenes determined. The limit of detection for a range of trichothecenes varied from 5 μg/kg to 15 μg/kg.