We complement the conditional capital asset pricing model (CAPM) by introducing unobservable long-run changes in risk factor loadings. In this environment, investors rationally “learn” the long-run level of factor loading by observing realized returns. As a direct consequence of this assumption, conditional betas are modeled using the Kalman filter. Because of its focus on low-frequency variation in betas, our approach circumvents recent criticisms of the conditional CAPM. When tested on portfolios sorted by size and book-to-market ratio, our learning-augmented conditional CAPM fails to be rejected. ; Original title: Learning about beta: a new look at CAPM tests.