A number of key insights into appropriate climate change policies have emerged from the application of integrated assessment models. These insights have been particularly powerful in those instances when they are shared by all or most of the existing integrated assessment models. On the other hand, some results and policy recommendations obtained from integrated assessment models vary widely from model to model. This can limit their usability for policy analysis. The differences between model results are mostly due to underlying assumptions about exogenous processes, about endogenous processes and the dynamics among them, differences in value judgments, and different approaches for simplifying model structure for computational purposes. Uncertainty analyses should be performed for the duel purpose of clarifying the uncertainties inherent in model results and improving decision making under uncertainty. This paper categorizes types of uncertainty analyses that can be performed on integrated assessment models. It then develops a unifying framework for comparing the different types of uncertainty analyses through their objective functions. The appendix presents a summary of integrated assessment models that explicitly account for uncertainty.