In spite of the increased use of factor-augmented regressions in recent years, little is known regarding the relative merits of the two main approaches to estimation and inference, namely, the cross-sectional average and principal component estimators. By providing a formal comparison of the approaches, the current paper fills this gap in the literature.