Empirical models of retrospective voting primarily employ standard monetary and financial indicators to proxy for voters' utility and to explain voters' behavior. We show that subjective well-being explains variation in voting intention that goes beyond what is captured by these monetary and financial indicators. For example, individuals who are satisfied with their life are 1.6% more likely to support the incumbent; by contrast, a 10% increase in family income leads to a 0.18% increase in an individual's support of the incumbent. We use difference-in-differences analysis to identify how voter intention is affected by a negative shock to well-being: the death of a spouse. Individuals who experience the death of a spouse are around 10% less likely than those in the control group to support the incumbent. The results hold even if elected officials' policies (health care, social welfare) cannot reasonably be blamed for the death.