The seller's risk-the probability of a set of samples exceeding an agreed u
pon aflatoxin level when the lot mean does not-and the buyer's risk-the pro
bability of a lot exceeding this level when a set of samples do not-have be
en computed using a parametrized experimental aflatoxin distribution and Mo
nte Carlo simulation. The calculations are exemplified using the proposed E
C standards (three 10 kg samples, 4 ng/g of total aflatoxin, basis kernels
only) as well as for samples up to 250 kg and for varied lot aflatoxin leve
ls. It is found that within this sample size range the seller's risk is as
high as 42% at 10 kg and increases with increasing sample size to 80% at 25
0 kg. Only by reducing lot levels to 0.2 ng/g of total aflatoxin, basis ker
nels, can the risk be brought down to 2.5%, independent of sample size. The
buyer's risk is as high as 58% at 10 kg but falls to 11% at 250 kg samp:le
s. The implications for both seller and buyer strategies are discussed.