Despite growing enthusiasm, there is little empirical evidence on how well energy efficiency investments work. Evidence is particularly lacking from low- and middle-income countries, despite a widespread view that these countries have many of the best opportunities. This paper evaluates a field experiment in Mexico in which a quasi-experimental sample of new homes was provided with insulation and other energy-efficient upgrades. A novel feature of our study is that we deploy large numbers of data loggers which allow us to measure temperature and humidity at high frequency inside homes. We find that the upgrades had no detectable impact on electricity use or thermal comfort, with essentially identical temperature and humidity levels in upgraded and non-upgraded homes. These results stand in sharp contrast to the engineering estimates that predicted up to a 26% decrease in electricity use. Part of the explanation is that air conditioning ownership is lower than expected, thus reducing the potential for reductions in energy use. In addition, we document that most households have their windows open on hot days, nullifying the thermal benefits of roof and wall insulation. Overall, we conclude that the benefits from these investments are unlikely to exceed costs, which added $400-$500 USD to the cost of each home. Our results underscore the urgent need to fully incorporate socioeconomic conditions and human behavior into engineering models of energy use.