We examine the evolution of the mechanism behind subjective well-being (SWB) across the lifespan using data from the 1972–2010 waves of the U.S. General Social Survey. By estimating a semiparametric varying-coefficient partially linear ordered probit model, we find that the influence of race, income, reference income, labor market status, marriage, and number of children on SWB varies greatly and nonlinearly along the life cycle. Among our results, we find that the effect of being black on the representative male's probability of being in the lowest happiness category falls from 9.399 percentage points (pps) at age 29 to 1.709 pps at age 55, turning insignificant afterwards. Being unemployed is associated with an increase in the representative male's probability of being in the lowest happiness category by 6.322 to 15.896 pps, with the largest effect occurring at age 40. The varying-coefficient model enhances our understanding of when life events are most detrimental to a person's well-being. The heterogeneity found highlights that, in order to promote well-being effectively, public policies should be differentiated across people depending on their age.