This study examined risk factors for posttraumatic stress disorder (PTSD) in Vietnam veterans: 68 women and 414 men of whom 88 were White, 63 Black, 80 Hispanic, 90 Native Hawaiian, and 93 Japanese American. Continuation ratio logistic regression was used to compare the predictive power of risk factors for the development versus maintenance of full or partial PTSD. The development of PTSD was related to premilitary, military, and postmilitary factors. The maintenance of PTSD was related primarily to military and postmilitary factors. Multivariate analyses identified different models for development and maintenance. We conclude that development of PTSD is related to factors that occur before, during, and after a traumatic event, whereas failure to recover is related primarily to factors that occur during and after the event.KEY WORDS: posttraumatic stress disorder; military veterans; longitudinal course; risk factors.Risk of developing posttraumatic stress disorder (PTSD) varies with a number of pretraumatic, peritraumatic, and posttraumatic factors (e.g., Breslau, Davis, Andreski, & Peterson, 1991;Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995;Resnick, Kilpatrick, Dansky, Saunders, & Best, 1993;Schnurr, Friedman, & Rosenberg, 1993). Risk of chronic PTSD, as reflected by a diagnosis of current PTSD among individuals whose exposure occurred many years prior, also is related to many of the same risk factors (e.g., King, King, Fairbank, Keane, & Adams, 1998;King, King, Foy, & Gudanowski, 1996;Kulka et al., 1990;Schnurr & Vielhauer, 1999;Stein, Walker, Hazen, & Forde, 1997). Yet very few investigations of risk factors for PTSD have attempted to distinguish chronic from less chronic PTSD. In fact, a recent meta-analysis combined studies of current and lifetime PTSD because of this limitation of the existing literature (Brewin, Andrews, & Valentine, 2000). Consequently, we know little about which factors are associated with the development of PTSD, as compared with the maintenance of PTSD. Such information has important implications for both theory and treatment.Studies of risk factors for lifetime PTSD yield information about the development of PTSD. Studies of risk factors for current chronic PTSD can yield information about maintenance if a comparison group without current PTSD is restricted to individuals who had PTSD in the past only. (If a comparison group also includes individuals who never had PTSD, risk factors for development are confounded with risk factors for maintenance.) Thus, comparisons between studies of risk factors for lifetime PTSD and studies of risk factors for current chronic PTSD can suggest which factors are associated with development and which are associated with maintenance. A limitation of this approach is that these comparisons are indirect.Another indirect approach is to examine the relative strength of predictors at different times. For example, Dunmore, Clark, and Ehlers (2001) studied assault survivors within 4 months after an assault, and then 6 and 9 months after the initial asses...