Background The purpose of this study was to identify latent classes of the severity of post-traumatic stress disorder (PTSD) among the survivors of the earthquake in Bam, south-eastern Iran, 17 years after the disaster. The most influential predictor variables of PTSD classes were also investigated. Methods Eight hundred survivors of the Bam earthquake who were at least four years old in that disaster were selected by multistage sampling. The PTSD Checklist-Civilian Version was used. Latent class analysis was performed to identify subgroups of people with different PTSD symptom profiles. Latent class regression analysis was used to explore the influence of demographic and traumatic variables on each class membership. Results We found three latent classes of PTSD, with the following profiles emerging: Low Symptom (56.6% of the participants), Moderate Symptom (23.5%), and Severe Symptom (19.9%). Old age [OR = 2.20, 95% CI = (1.46, 3.32)], physical injury [OR = 1.95, 95% CI = (1.24, 3.06)], being trapped under the rubble [OR = 1.81, 95% CI = (1.15, 2.86)], and death of a family member [OR = 1.86, 95% CI = (1.12, 3.09)] were positive risk factors for PTSD and increased the chance of being in classes having more severe symptoms. Having a high educational level was a negative risk factor [OR = 0.86, 95% CI = (0.67, 1.11)]. Conclusions The severity of PTSD 17 years after the earthquake shows that natural disasters such as earthquakes have long-term consequences, and earthquake survivors must have psychological support and long term health care. After any catastrophic earthquake, governments should establish psychology and psychotherapy centers for earthquake victims, and these centers should support earthquake victims for a sufficiently long time.