We exploit weekly options on the S&P 500 index to compute the weekly implied variance. We show that the weekly implied variance is a strong predictor of the weekly realized variance. In an encompassing regression test, it crowds out the information content of the monthly implied variance. Further tests reveal that the weekly implied variance outperforms not only the monthly implied variance but also well-established time series models of realized variance. This result holds both in- and out-of-sample and the forecast accuracy gains are significant.