Using data on the distributions of disease and pest resistance among wheat varieties, this paper reports theoretical and empirical findings concerning the efficient management of a gene bank and the optimal size of a collection.