The primary rationale for daylight saving time (DST) has long been energy savings. Whether it achieves this goal, however, remains a subject of debate. Recent studies, examining only one location at a time, have shown DST to increase, decrease or leave overall energy use unchanged. Rather than concluding the effect is ambiguous, this paper is the first to test for heterogeneous regional effects based on differences in (natural) sun times and (societal) waking hours. Using a rich hourly data set and quasi-experimental methods applied across Canadian provinces, this paper rationalizes the differing results, finding region-specific effects consistent with differences in sun times and waking hours. DST increases electricity use in regions with late sunrises and early waking hours.