Changes in climate, as well as direct effects of rising CO2 as a photosynthetic substrate, are inducing changes in the distribution of plant species, as well as their physiological performance, ecological relations (competition, pollination, disease occurrence), and evolution. Prediction of such changes is clearly important for ecology, as also for climatology, via vegetation feedbacks to the carbon, nutrient, and water cycles and regional albedos. Current dynamic vegetation models use climatic responses of vegetation, commonly statistical, with a modest inclusion of physiological responsiveness. The climatic models fail to account for abiotic (and biotic) extreme events, which are poorly characterized, if at all, by gross climatic averages over time and space. Moreover, novel climates are likely to arise, for which there is no basis to estimate plant responses. Rarely included in the models are the direct effects of CO2 on physiological performance, particularly resource-use efficiencies. Neither do models include the full, striking diversity of direct CO2 responses among species. Dynamic models are also limited in their capability to represent time lags, on annual to multi-decadal scales, arising from dispersal dynamics and from rapid changes in community structure in pulse-dominated ecosystems. Finally, there is dramatic evidence of constraint in vegetation responses at the population level and above, as a result of population genetic structure. We review the evidence for vegetation changes, the challenges in proper attribution of vegetation changes, and the capabilities of models, current and future.