The article explores the role of generalized or social trust (trust between strangers) in explaining cross-country differences in the level of productivity (output per worker) and in self-reported well-being for 136 countries. Trust is measured directly from survey data. In addition, a second trust variable called deep trust is estimated as a function of ancient cultural, historical, geographic, and linguistic factors. Both trust variables have significant bivariate relationships with each of productivity and well-being, each of which can also be modelled with fairly standard specifications: an augmented production function for productivity, and the multi-variate model of well-being developed in the annual World Happiness Reports. Yet when either trust variable is added to each of the standard models, neither contributes any additional explanatory power. So where is the bivariate significance of trust coming from? We find that, in every case, one or both of trust and deep trust is significant for the standard determinants of productivity, with deep trust doing better at predicting human capital, physical capital and institution quality, and actual trust being stronger for the well-being determinants. That is, trust in the 21st century appears to not directly contribute to productivity or well-being, but has a substantial effect working through the proximate determinants.