Considerable progress has been made in the past four years in defining a range of problems for which techniques are available to study impacts. The field, however, is waiting for improvements in our ability to foresee both the regional manifestations of a human-induced climate change and the change in variability of climate in each region. Farther down the road is the requirement that both modelers and impact experts learn to consider surprises that would move both the climate and the resulting impacts well outside the ranges now being considered.