A symposium discussion of the possible rates of genetic improvement of dairy cattle which seem possible with current knowledge. Rates of from 0.5-1% per year seem possible using only natural service and culling cows only on their own records and saving bulls only on the records of their dams. However, this requires that all selection be directed toward production. The rates can probably be made 1.5 times this high and possibly twice as high if optimum use is also made of sib tests, progeny tests and artificial breeding. New knowledge which seems needed if the rates are to be increased much, if at all, beyond these levels is discussed.