Using a repeated-measures design, we found a significant birth-order relationship suggesting lower ages of smoking onset in later born siblings of a 1979 National Longitudinal Survey of Youth cohort. Two social learning mechanisms, modeling and opportunity, were explored to help illuminate the causes of trends in the within-family means. When empirical patterns were compared to predictions derived from our specifications of how opportunity and modeling processes should work, the results were unsuccessful in explaining the birth-order effect. As a third explanation of the birth-order effect, telescoping did show a significant influence. The effect size was small, however, and had little effect on the group means assessed. Finally, a pattern did emerge that was consistent with a reformulation of the opportunity process in which sisters play a particularly strong role. We develop future research implications of this pattern and speculate on genetic and social conservatism explanations.