Recent events suggest that uncertainty changes play a major role in U.S. labor market fluctuations. This study analyzes the impact of uncertainty shocks on unemployment dynamics. Using a vector autoregression approach, we show that uncertainty shocks measured by stock market volatility have a significant impact on the U.S. unemployment rate. We then develop a quantitative version of the Diamond-Mortensen-Pissarides (DMP) model, in which uncertainty shocks hit the economy. Given the significant nonlinearities of the DMP model, we show that the introduction of uncertainty shocks not only allows this textbook model to account for observed characteristics of the U.S. labor market dynamics, with reasonable values for calibrated parameters, but also for the impact of rare episodes such as economic crises.