When agents devise plans for execution in the real world, they face two forms of uncertainty" they can never have complete knowledge about the state of the world, and they do not have complete control, as the effects of their actions are uncertain. Most of the approaches for planning under uncertainty avoid explicit uncertain reasoning by devising methods to render plans less rigid and less dependent on the exact state of the world. We believe that a planning framework should explicitly represent uncertainty. We develop a probabilistic representation for states and events, based on belief nets. We define conditional belief nets (CBNs) to capture the intrinsic relationships among entities in the environment, and the probabilistic dependency of the effects of an event upon the state of the world. CBNs can be interpreted to represent causal relations in a more profound way than belief nets. We present a simple projection algorithm to construct the belief net of the state succeeding an event. We discuss how the qualitative aspects of belief nets and CBNs make them more appropriate than other probabilistic representations for the various stages of the problem solving process, from model construction to the design of planning algorithms.