Few studies have investigated the patterns of posttraumatic stress disorder (PTSD) symptom change in prolonged exposure (PE) therapy. In this study, we aimed to understand the patterns of PTSD symptom change in both PE and present-centered therapy (PCT). Participants were active duty military personnel (N = 326, 89.3% male, 61.2% white, 32.5 years old) randomized to spaced-PE (S-PE; 10 sessions over 8 weeks), PCT (10 sessions over 8 weeks), or massed-PE (M-PE; 10 sessions over 2 weeks). Using latent profile analysis, we determined the optimal number of PTSD symptom change classes over time and analyzed whether baseline and follow-up variables were associated with class membership. Five classes, namely rapid responder (7-17%), steep linear responder (14-22%), gradual responder (30-34%), non-responder (27-33%), and symptom exacerbation (7-13%) classes, characterized each treatment. No baseline clinical characteristics predicted class membership for S-PE and M-PE; in PCT, more negative baseline trauma cognitions predicted membership in the non-responder v. gradual responder class. Class membership was robustly associated with PTSD, trauma cognitions, and depression up to 6 months after treatment for both S-PE and M-PE but not for PCT. Distinct profiles of treatment response emerged that were similar across interventions. By and large, no baseline variables predicted responder class. Responder status was a strong predictor of future symptom severity for PE, whereas response to PCT was not as strongly associated with future symptoms.