In a longitudinal study we assessed which baseline risk factors are associated with persistent and partially remitted PTSD in comparison to fully remitted PTSD. 6427 (68%) of a randomly selected sample of UK service personnel completed the PTSD checklist (PCL) between 2004 and 2006 (Phase 1) and between 2007 and 2009 (Phase 2). 230 (3.9%) had possible PTSD at baseline. 66% of those with possible PTSD at baseline remitted (PCL score <30) or partially remitted (PCL score 30-49) by phase 2 of the study. Associations of persistent PTSD with the fully remitted group for risk factors at phase 1 adjusted for confounders were having discharged from service (OR 2.97, 95% Cl 1.26-6.99), higher educational qualification (OR 2.74, 95% 1.23-6.08), feeling unsupported on return from deployment (OR 10.97, 95% Cl 3.13-38.45), deployed not with parent unit (OR 5.63, 95% Cl 1.45-21.85), multiple physical symptoms (OR 3.36, 95% Cl 1.44-7.82), perception of poor or fair health (OR 2.84, 95% Cl 1.28-6.27), older age and perception of risk to self (increasing with the number of events reported, p = 0.04). Deploying but not with a parent unit and psychological distress were associated in the partially remitted PTSD when compared to the fully remitted group. The positive and negative likelihood ratios for the factors most highly associated with persistent PTSD indicated they were of marginal value to identify those whose presumed PTSD would be persistent. Many factors contribute to the persistence of PTSD but none alone is useful for clinical prediction. (C) 2012 Elsevier Ltd. All rights reserved.