This article examines three hypotheses about the relation between age and the stability of sociopolitical attitudes. The hypotheses are (1) the impressionable-year hypothesis, which states that the youngest adults have the least stable attitudes; (2) the aging-stability hypothesis, that attitude stability increases with age; and (3) the hypothesis that symbolic attitudes are more likely to show distinctive life-cycle patterns of attitude stability than less symbolic ones. The hypotheses are tested using nationally representative panel data from the National Election Study (NES). When results are aggregated over 50 different measures of attitudes, they reveal that in general the youngest adults have the lowest levels of attitude stability, although the difference is not significant. Beyond this, the aggregated data show very few systematic age-related differences, and very few life-span differences in attitude stability are related to the nature of the attitude object; that is, symbolic attitudes do not seem to differ systematically from nonsymbolic attitudes in the relationship of age to stability. However, the examination of intracohort patterns of change in stability, using a comparison of stabilities in political party identification across the 1956-1958-1960 and the 1972-1974-1976 NES panel studies, reveals systematic differences that provide clear support for the impressionable-years and aging-stability hypotheses. The decomposition of the stabilities in this measure into components representing "direction" and "intensity" of political partisanship suggests that the intensity component of partisan attitudes declines in stability in old age, whereas the stability of the direction of party loyalties either increases or persists with age. The prevailing model of political socialization-that persons become more "persistent" with age-is reevaluated on the basis of these findings.