This paper describes a model for estimating, in a probabilistic framework, expected future consumer surplus from planned new product innovations. The model has been applied to estimations of taxpayer benefits from NASA's New Millenium Program (NMP), which develops new technologies for space science, and to the digital data storage technologies being supported by the Department of Commerce's Advanced Technology Program (ATP). The model uses cost index methods based on consumers' estimated marginal valuation for quality improvements in the technology. Probabilistic values for performance increases are taken from the innovators' own expectations. The analysis reveals the sensitivity of welfare increases to these values, which are assumed to be biased upward. The cost index, when combined with an expected rate of adoption, estimates consumer benefits from the innovation, gross of its research and development costs. Benefits are estimated net of a dynamic baseline defined by the best available substitute technology, which is also assumed to improve over time. Findings are therefore expressed in terms of the economic value of the innovation to consumers, compared to advances that might occur in the absence of the NMP or ATP investments. Illustrative results--estimated cost indices and 95% confidence bounds--are presented for technologies that are expected to improve consumer welfare and for those that, on a quality-adjusted cost basis, are likely to be outperformed by the selected baseline technology.