We highlight the limitations of using cumulative advertising expenses as an input measure of brand value. Using two output measures—Interbrand’s data and a novel text-based measure—we find that an equal-weighted portfolio of top brands yields an annual abnormal return of 3%. The excess returns are driven by companies that develop their brands internally. Intangible factors proposed in the literature have no explanatory power for the premium. Analysts underestimate the future earnings of top brands, leading to significant excess returns following earnings announcements. We find no abnormal returns associated with the input measure of brand value.