Cognitive processing therapy (CPT) for posttraumatic stress disorder (PTSD) has been shown to reduce symptoms of PTSD in a veteran population. This study explored patterns of self-reported symptom change during CPT. Veterans (N = 60) with PTSD were randomized to receive CPT immediately or after 10 weeks. We hypothesized that those treated immediately would evidence initial symptom stability followed by decline compared with those who waited, whose PTSD symptoms would remain stable. The best model fit based on deviance statistics and Bayesian information criteria comparisons was one in which participants treated immediately showed more rapid initial decline followed by a slower rate of PTSD symptom improvement relative to those who waited, who showed a stable level of symptomatology. Findings suggest that CPT produces quick and maintained improvements in veterans. The effect sizes for change between those who received CPT immediately and those who waited were approximately medium sized. Implications of findings are discussed.