The stress-acceleration hypothesis of nightmares (Nielsen, 2017) stipulates that individuals with frequent nightmares have better access to memories—including dreams—originating in the infantile amnesia period than do individuals without nightmares. This was tested on an available sample of 17,014 participants who estimated their current nightmare frequency and dated their earliest remembered dreams. One-way analyses of variance with 10 levels of Dream-Age (1–10 years) as independent variable and log nightmare recall as the dependent measure were computed for all early dreams combined and, separately, for those who remembered only positive or negative early dreams. An earliest dream from the infantile amnesia period was recalled by 4.63% of participants. A main effect for Dream-Age (p < .0000001) confirmed that these participants had more current nightmares. The effect was also seen for both emotionally positive and negative early dreams, suggesting a general change in early memory access. Four themes accounted for most (40.2%) of the earliest dreams recalled—being chased (14.0%), falling (11.4%), flying (9.4%), and encountering an evil force (5.3%)—and were interpretable as consistent with Freud’s claim that such dreams arise from early infantile experiences. Results support the stress-acceleration hypothesis of nightmares, which stipulates that a foreshortening of the infantile amnesia period contributes to nightmares by increasing access to intense primordial feelings and memory fragments that are normally forgotten because of infantile amnesia processes.