The significance of laboratory sample preparation for the determination of two important mycotoxins, ochratoxin A (OTA) and deoxynivalenol (DON), in wheat was investigated by comparing water-slurry mixing and dry-milling procedures. The distribution of OTA and DON in 10 kg samples of naturally contaminated wheat was established by analyzing one hundred 100 g subsamples of each sample. A normal distribution and a good repeatability of DON measurements was observed for both water-slurry mixing (mean 2290 microg/kg, CV 4.6%, median 2290 microg/kg) and dry milling (mean 2310 microg/kg, CV 6.4%, median 2290 microg/kg) procedures. For OTA determinations, reliable results could be obtained only by slurry mixing sample preparation (mean 2.62 microg/kg, CV 4.0%, median 2.62 microg/kg), whereas dry-milling comminution resulted in an inhomogeneous distribution with a high variability (mean 0.83 microg/kg, CV 75.2%, median 0.60 microg/kg) and a positive skewness (2.12). Ad hoc experiments were performed on different size portions of the same sample (10 kg) to assess accuracy and precision of the comminution/homogenization procedures (slurry mixing and dry milling). Very good results were obtained for DON determination with both procedures in terms of accuracy (>98.7% of the "weighted value") and precision (CV <3%). For OTA determination good results were only obtained by slurry mixing (99.4% of the "weighted value," CV 10%), whereas dry milling provided results with low accuracy (43.2% of the "weighted value") and high variability (CV 110%). This study clearly demonstrated that sample preparation by slurry mixing is strictly necessary to obtain reliable laboratory samples for OTA determination in wheat to minimize misclassification of acceptable/rejectable lots, mainly within official control.