Background: Early life stress (ELS) affects facial emotion recognition (FER), as well as the underlying brain network. However, there is considerable inter-individual variability in these ELS-caused alterations. As the hypothalamic-pituitary-adrenal (HPA) axis is assumed to mediate neural and behavioural sequelae of ELS, the genetic disposition towards HPA axis reactivity might explain differential vulnerabilities. Methods: An additive genetic profile score (GPS) of HPA axis reactivity was built from 6 SNPs in 3 HPA axisrelated genes (FKBP5, CRHR1, NR3C1). We studied two independent samples. As a proof of concept, GPS was tested as a predictor of cortisol increase to a psychosocial challenge (MIST) in a healthy community sample of n=40. For the main study, a sample of n=170 completed a video-based FER task and retrospectively reported ELS experiences in the Childhood Trauma Questionnaire (CTQ). Results: GPS positively predicted cortisol increase in the stress challenge over and above covariates. CTQ and genetic profile scores interacted to predict facial emotion recognition, such that ELS had a detrimental effect on emotion processing only in individuals with higher GPS. Post-hoc moderation analyses revealed that, while a less stress-responsive genetic profile was protective against ELS effects, individuals carrying a moderate to high GPS were affected by ELS in their ability to infer emotion from facial expressions. Discussion: These results suggest that a biologically informed genetic profile score can capture the genetic disposition to HPA axis reactivity and moderates the influence of early environmental factors on facial emotion recognition. Further research should investigate the neural mechanisms underlying this moderation. The GPS used here might prove a powerful tool for studying inter-individual differences in vulnerability to early life stress.