This prospective longitudinal study examined the ability of re-experiencing, avoidance, numbing, and hyperarousal symptoms to predict persistence of posttraumatic stress disorder (PTSD) in disaster workers followed for 2 years. Cluster analyses suggested that overall severity was the best predictor of PTSD at follow up, but for groups with PTSD of moderate severity, numbing symptoms were also associated with PTSD at the 2-year follow up. Regression analyses with all four symptom groups as independent variables found that only numbing and re-experiencing symptoms predicted PTSD at the 1 year follow up, and only numbing symptoms predicted PTSD at the 2-year follow up. Findings suggest that numbing symptom severity could be used as a risk index of very chronic PTSD, especially when the overall PTSD severity falls in the moderate range.