Primary dysmenorrhea (PDM) refers to menstrual pain of which the pathological cause(s) are unknown. This study examined the associations among BDNF Val66Met polymorphisms, menstrual pain severity, and hippocampal volume among young PDM subjects. We recruited 115 PDM subjects, including severe cases (n = 66) and moderate cases (n = 44), and 117 young females (aged 20-30 years) as a control group (CON) for BDNF Val66Met genotyping and MRI examination. The assessment of hippocampal volume involved analysis at various anatomical resolutions, i.e., whole hippocampal volume, hippocampal subfields, and voxel-based morphometry (VBM) volumetric analysis. Two-way ANOVA analyses with planned contrasts and Bonferroni correction were conducted for the assessment of hippocampal volume. Linear regression was used to test for BDNF Val66Met Val allele dosage-dependent effects. We observed no main effects of group, genotype, or group-genotype interactions on bilateral whole hippocampal volumes. Significant interactions between PDM severity and BDNF Val66Met genotype were observed in the right whole hippocampus, subiculum, and molecular layer. Post-hoc analysis revealed that the average hippocampal volume of Val/Val moderate PDM subjects was greater than that of Val/Val severe PDM subjects. Note that right hippocampal volume was greater in the Val/Val group than in the Met/Met group, particularly in the right posterior hippocampal region. Dosage effect analysis revealed a positive dosage-dependent relationship between the Val allele and volume of the right whole hippocampus, subiculum, molecular layer, and VBM-defined right posterior hippocampal region in the moderate PDM subgroup only. These findings indicate that Val/Val PDM subjects are resistant to intermittent moderate pain-related stress, whereas Met carrier PDM subjects are susceptible. When confronted with years of repeated PDM stress, the hippocampus can undergo differential structural changes in accordance with the BDNF genotype and pain severity. This triad study on PDM (i.e., combining genotype with endophenotype imaging results and clinical phenotypes), underscores the potential neurobiological consequences of PDM, which may prefigure in neuroimaging abnormalities associated with various chronic pain disorders. Our results provide evidence for Val allele dosage-dependent protective effects on the hippocampal structure; however, in cases of the Val variant, these effects were modulated in accordance with the severity of menstrual pain.