In this study, we provide empirical evidence that firm-level jump-induced tail risk (measured by a jump-implied variance contribution index [JIVX]) prospectively predicts cross-sectional stock returns around earnings announcements. The effect size is nontrivial. A practical trading strategy that buys announcers with high pre-news JIVX values and sells announcers with low pre-news JIVX values, earns a net risk-adjusted average return of 82 basis points (bps) three days after the news release. Notably, the empirical success of the JIVX predictor is distinct from model-free implied skewness and kurtosis measures and withstands a battery of robustness checks.