Rising energy prices and environmental concerns are driving countries to consider extending Daylight Saving Time (DST) in order to conserve energy. Beginning in 2007, the U.S. will lengthen DST by one month with the specific goal of reducing electricity consumption by 1%. In this paper we question the findings of prior DST studies, which often rely on simulation models and extrapolation rather than empirical evidence. By contrast, our research exploits a quasi-experiment, in which parts of Australia extended DST by two months to facilitate the Sydney Olympic Games in 2000. Using detailed panel data on half-hourly electricity consumption, prices, and weather conditions, we show that the extension failed to reduce electricity demand. We further examine prior DST studies and find that the most sophisticated simulation model available in the literature significantly overstates electricity savings when it is applied to the Australian data. These results suggest that current plans and proposals to extend DST will fail to conserve energy.