To understand the development of crown structure in Betula platyphylla Sukatch., mortality patterns of long shoots were analyzed quantitatively. We selected 25 saplings growing under various light conditions and measured the relative photosynthetically active radiation (WAR) at, and the three-dimensional position of, first-order branches. A long shoot was assigned "no buds" (NB) status if it lacked buds at the end of the growing season, including at the tips of short shoots. A long shoot was classified as dead if it was NB and all the offspring long shoots issuing from it were NB. The probability that a leafy long shoot (a current-year long shoot with leaves or an older long shoot with short shoots with leaves) would become NB by the end of the season was positively dependent on shoot age and branch age, and negatively dependent on shoot length, centripetal shoot order, branch height and WAR at the branch. Randomization tests revealed that shoots became NB and dead in clusters of connected shoots. In particular, shoot clusters originating from 3-year-old shoots were more likely to die than expected if each shoot was assumed to become NB regardless of the connection. Stepwise logistic regression revealed that the maximum WAR within the crown of an individual tree had a significant effect on the mortality rate of 3-year-old shoot clusters, together with the WAR at the level of the branch and other structural entities. Correlative inhibition is an important mechanism for determining shoot mortality patterns.