Given the high prevalence of early life stress (ELS) and the potential physiological dysregulation such experiences can lead to, this meta-analysis tested the relationship between ELS and cortisol. Search terms related to ELS and cortisol were entered in to PsycINFO and PubMed. Effect sizes were extracted for four outcomes variables: cortisol awakening response (CAR), baseline cortisol (cortisol at one time point), non-stressed cortisol over time (cortisol captured at two or more time points), and cortisol reactivity to an acute stressor. The articles were additionally coded for potential confounding variables, population-related, ELS-related and cortisol-related moderators. There was no significant relationship between ELS and the CAR (g = 0.19, p = 0.268), ELS and baseline cortisol (g = − 0.072, p = 0.328), ELS and non-stressed cortisol over time (g = 0.09, p = 0.292) or ELS and cortisol reactivity (g = − 0.089, p = 0.363). However, there was a significant amount of heterogeneity amongst relationships. Within the ELS-CAR relationship, in those who had experienced ELS that was sexually, physically or emotionally abusive, the CAR was heightened. Within the ELS-Baseline relationship, if blood samples were collected the ELS was associated with a blunting effect of cortisol. The non-significant main effects challenge the commonly held belief in the literature that ELS affects cortisol later in life. However, the high degree of heterogeneity uncovered by this analysis and significant moderators suggest that the literature may benefit from consistent operationalizations of ELS and standardized methods of how cortisol is measured.