The additive and non-additive variance components were estimated from progenies derived from two samples of parents (representing a northern continental type climate) for five factors relating to plant survival and two composites of the factors. It was found that additive variance made up 90 and 100%, 91 and 100%, 91 and 100%, 100 and 100%, 82 and 59%, 91 and 100%, and 90 and 100% of the total genetic variance for leafing-out date, leafingout percent, tip injury, stem damage, root damage, a shoot composite, and a shoot-root composite for the two samples respectively. A third sample had 100% additive variance for plant height while, in contrast, a sample of rootstocks, differing from each other in their ability to dwarf grafted scions, had approximately 50-70% additive variance for plant height. It was shown that breeding progress for both winter survival and plant height could be achieved by exploiting the additive variance, the total genetic variance, or (where progenies were the selection unit rather than individuals) by progeny selection. By exploiting the additive variance, it should be possible to improve plant survival and change plant height in each of several successive generations. It is predicted that (with the exception of selection for vigor in a population having a range of dwarfing abilities) potential parents could be efficiently screened phenotypically and so obviate the need for genotypic evaluation. A total of 9180 progeny trees were involved in the analyses considered in this paper.