The timing of spring bud-burst and leaf development in temperate, boreal and Arctic trees and shrubs fluctuates from year to year, depending on meteorological conditions. Over several generations, the sensitivity of bud-burst to meteorological conditions is subject to selection pressure. The timing of spring bud-burst is considered to be under opposing evolutionary pressures; earlier bud-burst increases the available growing season (capacity adaptation) but later bud-burst decreases the risk of frost damage to actively growing parts (survival adaptation). The optimum trade-off between these two forms of adaptation may be considered an evolutionarily stable strategy that maximizes the long-term ecological fitness of a phenotype under a given climate. Rapid changes in climate, as predicted for this century, are likely to exceed the rate at which trees and shrubs can adapt through evolution or migration. Therefore the response of spring phenology will depend not only on future climatic conditions but also on the limits imposed by adaptation to current and historical climate. Using a dataset of bud-burst dates from twenty-nine sites in Finland for downy birch (Betula pubescens Ehrh.), we parameterize a simple thermal time bud-burst model in which the critical temperature threshold for bud-burst is a function of recent historical climatic conditions and reflects a trade-off between capacity and survival adaptation. We validate this approach with independent data from eight independent sites outside Finland, and use the parameterized model to predict the response of bud-burst to future climate scenarios in north-west Europe. Current strategies for budburst are predicted to be suboptimal for future climates, with bud-burst generally occurring earlier than the optimal strategy. Nevertheless, exposure to frost risk is predicted to decrease slightly and the growing season is predicted to increase considerably across most of the region. However, in high-altitude maritime regions exposure to frost risk following bud-burst is predicted to increase.