Data from 110 U.S. national surveys are used to study the trend in “no religion” responses from the late 1950s to the early 1980s. The trend was monotonically upward, at least until the 1980s, when there were signs of leveling off or an incipient reversal of the trend. The trend came about through both changes within birth cohorts and cohort succession, and it involved persons of all major religious backgrounds and both men and women. Whether or not the change indicates an important degree of secularization is considered at length. Problems involved in using a dichotomous indicator to gauge change in a quantitative variable are discussed.