Earth-science predictions of natural phenomena are increasingly seen as valuable aids to improved societal decision making. Pielke et al. recently (EOS 7/13/99) argued persuasively that good predictions alone won’t achieve better societal decisions. These authors’ call to change the decision environments in which scientific predictions are used, though, may be more relevant to the daily activities of policy makers than to those of scientists. We see a role also for changing the information that scientists feed into those decision environments. In particular, scientists could better serve societal needs by generating not only possible scenarios, but also improved probabilities that decision makers need, including for decisions to be taken in the near future.