Objective: Posttraumatic stress disorder (PTSD) is prevalent and sometimes severely disabling. Providing effective treatment for PTSD and addressing its social consequences require accurate diagnosis. PTSD criteria have changed in all editions of the American Diagnostic Criteria since introduction of the diagnosis in DSM-III in 1980. The DSM-5 Field Trials demonstrated very good inter-rater reliability for PTSD, but a crosswalk study comparing DSM-IV and DSM-5 criteria has potential to identify diagnostic differences generated by changed criteria. Methods: A DSM-IV to DSM-5 PTSD crosswalk study was conducted in real-world adult clinical treatment settings in two DSM-5 Field Trials sites, the Dallas (N = 93) and Houston (N = 48) Veterans Affairs medical centers. The crosswalk assessment was conducted by trained clinicians who interviewed the patients and rated both sets of criteria on a combined checklist. Results: PTSD prevalence differed insubstantially between criteria sets (42% vs. 45% and 55% vs. 52% in the Dallas and Houston sites, respectively), with moderate to excellent diagnostic agreement (reliability indicated, respectively, by κ = .53 and .93); however, substantial proportions of individuals diagnosed in one criteria set did not meet criteria in the other. Differences in cross-criteria diagnostic reliability were largely a function of differing definitions of criterion A trauma. Conclusions: Reliability across the two criteria sets was generally good to excellent, and diagnostic discrepancy predominantly reflected the elimination of criterion A2 in DSM-5 with a smaller contribution from changes to the avoidance and numbing criteria.