Recent cross-section studies find that advertising is relatively long-lived, providing support for treating advertising as "capital." This paper models the firm's advertising decision, treating advertising and product quality as complements. The model predicts that failing to control for differences in firm-specific factors positively correlated with the return to advertising will understate advertising's direct effect on sales and overstate its durability. Our empirical study, using several years of data for over 400 firms, confirms these predictions: advertising's apparent life shrinks dramatically when firm-specific factors are held constant. We also find no support for the steady-state assumptions critical to intangible assets models.