Uncertainty faced by individual firms appears to be heterogeneous. In this paper, I construct new empirical measures of firm-level uncertainty using data from the I/B/E/S and Compustat. These new measures reveal persistent differences in the degree of uncertainty facing individual firms not reflected by existing measures.  Consistent with existing measures, I find that the average level of uncertainty across firms is countercyclical, and that it rose sharply at the start of the Great Recession. I next develop a heterogeneous firm model with Bayesian learning and uncertainty shocks to study the aggregate implications of my new empirical findings. My model establishes a close link between the rise in firms’ uncertainty at the start of a recession and the slow pace of subsequent recovery. These results are obtained in an environment that embeds Jovanovic’s (1982) model of learning in a setting where each firm gradually learns about its own productivity, and each occasionally experiences a shock forcing it to start learning afresh. Firms differ in their information; more informed firms have lower posterior variances in beliefs. An uncertainty shock is a rise in the probability that any given firm will lose its information. When calibrated to reproduce the level and cyclicality of my leading measure of firm-level uncertainty, the model generates a prolonged recession followed by anemic recovery in response to an uncertainty shock. When confronted with a rise in firm-level uncertainty consistent with advent of the Great Recession, it explains 79 percent of the observed decline in GDP and 89 percent of the fall in investment.