Costs and benefits of building energy efficiency are estimated as a means of reducing greenhouse gas emissions in Pittsburgh, PA and Austin, TX. The analysis includes electricity and natural gas consumption, covering 75% of building energy consumption in Pittsburgh and 85% in Austin. Two policy objectives were evaluated: maximize GHG reductions given initial budget constraints or maximize social savings given target GHG reductions. This approach evaluates the trade-offs between three primary and often conflicting program design parameters: initial capital constraints, social savings, and GHG reductions. Results suggest uncertainty in local stocks, demands, and efficiency significantly impacts anticipated outcomes. Annual GHG reductions of 1 ton CO2 eq/capita/yr in Pittsburgh could cost near nothing or over $20 per capita annually. Capital-constrained policies generate slightly less social savings (a present value of a few hundred dollars per capita) than policies that maximize social savings. However, sectors and end uses targeted for intervention vary depending on policy objectives and constraints. Optimal efficiency investment strategies for some end uses vary significantly (in excess of 100%) between Pittsburgh and Austin, suggesting that resources and guidance conducted at the national scale may mislead state and local decision-makers. Results are used to provide recommendations for efficiency program administrators. (C) 2011 Elsevier Ltd. All rights reserved.