Wheat scab, caused by Gibberella zeae, has been a serious disease in parts of the Midwest. One factor contributing to the importance of wheat scab is the contamination of grain by the mycotoxin vomitoxin (deoxynivalenol, DON), a toxic secondary metabolite. The U.S. Food and Drug Administration advisory levels for vomitoxin in wheat and wheat products require an accurate and precise assessment of vomitoxin concentration. In this study, randomly collecting probes of wheat from individual trucks and analyzing the ground wheat from each probe produced significantly less variability than subsampling and analyzing 50 g of whole kernels from the probes. The variability introduced by subsampling the probes and analyzing 50 g of whole kernels affects the precision and confidence of vomitoxin estimates. Tables of confidence intervals were developed for different sampling and subsampling patterns. To be 95% certain that the true vomitoxin concentration does not exceed the sample estimate by 1 μg/g, analyzing either four individual probes or 5--12% subsamples of these probes would be sufficient. To increase the accuracy to about 0.5 μg/g, either an analysis of seven probes or a 5--12% subsample of 10 probes would be necessary, based on a one-sided confidence interval.