To better quantify the effects of conservation measures, degree-day-based techniques are commonly used to isolate weather-induced changes in building energy use. In this paper, we use a building energy simulation model, which allows us to hold fixed all influences on energy use besides weather, to evaluate several degree-day-based techniques. The evaluation is applied to simulated electricity and natural gas consumption for two large office building prototypes located in five U.S. climates. We review the development of degree-day-based, weather-normalization techniques to identify issues for applying the techniques to office buildings and then evaluate the accuracy of the techniques with the simulated data. We conclude that, for the two office building prototypes and five U.S. locations examined, most techniques perform reasonably well; accuracy, in predicting annual consumption, is generally better than 10%. Our major finding is that accuracy among individual techniques is overwhelmed by circumstances outside the control of the analyst, namely, the choice of the initial year from which the normalization estimates are made.