Although the effects of deterministic factors on population viability often are more important than stochasticity, few researchers have dealt with the effect of deterministic habitat changes on plant population demography We assessed population viability for the perennial herb Primula veris L. and identified targets for management based on demographic data from five different habitat types representing different degrees of canopy closure. We conducted replicate studies at the border of the distribution area and in more central parts. Demographic patterns were similar between the two regions. Most study populations had a positive population growth, and only populations in late phases of forest succession showed consistently negative trends. The populations of open habitats had high seedling recruitment, and the populations of early and middle forest succession had high seed production. The importance of survival for population growth rate increased with increasing habitat closure, whereas the importance of growth and reproduction decreased. Results of the elasticity analysis suggested that the best method to manage decreasing late-successional populations is to increase survival of the largest individuals. The life-table response experiment (LTRE) analysis, however, showed that survival of the largest individuals contributed little to differences in population growth rates of different habitats, whereas seed production and growth of small individuals were more important. Moreover, direct perturbation of the performance of the largest stages showed that late-successional populations would not attain positive population growth even if the largest stages had no mortality at all. We conclude that restoration of recruitment is the only possibility for positive population growth in late-successional populations of P. veris, although the elasticities of recruitment transitions are low. Our results also suggest that retrospective demographic methods such as LIRE constitute an important and necessary complement to prospective methods such as elasticities in identifying management targets.