Following large-scale disasters, there is a well-documented increase in medically unexplained symptoms in survivors. This increase in somatization appears to be related to an individual's attachment style, social support, and degree of exposure to the disaster. However, few studies are able to longitudinally analyze such relationships because disasters are difficult to predict and sufficient data are rarely available from immediately before the disaster to allow researchers to assess the effects of pre-trauma psychological variables on reactions to the trauma. The present study investigates the effects of the April 27, 2011 tornado of Tuscaloosa, Alabama, using data collected from screening measures administered to University of Alabama Psychology subject pool participants 2 weeks to 8 months prior to the tornado, as well as follow-up measures collected approximately 6 to 8 months after the tornado. Multiple regression analyses and structural equation modeling were conducted to assess the effects of attachment style, social support, and degree of exposure to the disaster on changes in somatization following the tornado. Two analyses were conducted, one using pre-tornado attachment and one using post-tornado attachment. In both of these analyses, high disaster exposure and low social support significantly predicted increased somatization. Post-tornado insecure anxious attachment predicted increased somatization, and this effect was most pronounced in participants with high levels of storm exposures. However, these effects were not found for pre-tornado insecure anxious attachment. For both the pre-and post-tornado attachment models, all simple and moderated effects of attachment on increasing somatization were mediated by poor social support. While these results confirm the importance of disaster exposure and social support in predicting symptom change, the inconsistency between the prospective and cross-sectional findings related to attachment cast doubt on a straightforward view that insecure attachment is a risk factor for somatization.