This systematic review examines moderators and mediators tested in evaluations of stress management interventions for hospital employees to determine their significance for intervention outcomes. To be included, studies had to comprise a moderator or mediator analysis and a quantitative assessment of stress or mental well-being, and to be published in English or German language. Five databases (APA PsycInfo, APA PsycArticles, Embase, Medline, and Web of Science) were searched. Moderators and mediators were categorised thematically and examined using effect direction plots. Study quality was assessed using RoB 2 and ROBINS-I. In fifteen included studies, 22 moderators and ten mediators were identified. Moderators and mediators were categorised into individual psychological factors (14), socio-economic status (6), work situation (5), intervention (3), and duration of employment (3). Two moderators (perceived stressfulness of residency, job control) had a positive, two a negative impact (spirituality, socially desirable responding). One moderator (years of professional experience) had a positive and negative impact. Three moderators measured on categorical scales (gender, profession, and shiftwork) also had effects, favouring women, physicians and night-shift employees. Five mediators (adherence to intervention, mindfulness, non-reactivity to inner experience, total observing, and self-compassion) had a positive impact, while three (isolation, over-identification, psychological inflexibility) had a negative impact. In conclusion, effects of interventions were predominantly driven by individual psychological factors, while the role of other variables seems to be limited. Interventions focussing on primary or tertiary prevention were rare. Also processes through which organisational-level interventions can be most effective have been hardly investigated. Larger and methodologically robust studies are needed to better understand causal pathways and optimise matching of interventions to target groups.