High (low) quality stocks generate anomalously high (low) returns above and beyond expected returns based on betas, market sizes, valuations, and momentum. We provide a comprehensive overview of commonly used quality definitions and test their predictive power for stock returns. We show that quality measures predict stock returns if and only if they forecast earnings growth, and that this information is not contained in other characteristics that have been shown to drive expected stock returns. At the same time, we find that the quality premium is unrelated to different measures of distress risk, and therefore inconsistent with a risk-based interpretation. Finally, our results are robust across different regions and carry over to the corporate bond market.