The significant power optimization possibilities in the early stages of the design flow advice the use of energy evaluation techniques at high levels of abstraction. With this aim, the present work addresses the estimation of the energy consumption in very deep submicron technologies. Using the characterization of the probability density function with a projection in an orthogonal polynomial base, and a symbolic propagation mechanism, a technique is presented to estimate the dynamic and static power consumption in digital systems. The proposed approach has been validated with circuits and excitations from realistic applications. Comparisons with reference transistor and bit level simulations are reported in order to asses the the accuracy of the technique.