Studies were conducted in order to assess the use of binomial sampling for prediction of infestation level in stored wheat. In each of three steel silos with 1500 metric tones of wheat each, located in central Greece, 14 probe traps were placed on 15 June 1997. The traps were checked for adult coleoptera every 15 days, from 30 June until 30 January 1998. On the same dates, 14 wheat samples were taken adjacent to the trap locations, using a grain trier. Most abundant species were found to be Cryptolestes ferrugineux and Tribolium castaneum in the traps, while Sitophilus oryzae in the samples. Regarding all species detected, traps were proved to be more effective as compared to the samples. Taylor's Power Law was used, in order to estimate y-intercept and slope values for each species. The comparison of these parameters indicated that a single (weighted) equation can describe equally well the relation between the mean and the variance, according to Taylor's Power Law, for all adults found, regardless of species. The parameters of this relation were utilized to connect the ratio of sampling units containing one or more adults and the mean number of adults per sampling unit (x), using Wilson and Room's model. Regarding trap catches, the same model can be used to predict an infestation, with a sufficient precision level, mainly when x<5; on the contrary, the results were not satisfactory in the case of adult numbers in the samples.