Many workers suffer from work-related stress and are at increased risk of work-related cardiovascular, musculoskeletal, or mental disorders. In the European Union the prevalence of work-related stress was estimated at about 22%. There is consensus that stress, absenteeism, and well-being of employees can be influenced by leadership behaviour. Existing reviews predominantly included cross-sectional and non-experimental studies, which have limited informative value in deducing causal relationships between leadership interventions and health outcomes. To assess the effect of four types of human resource management (HRM) training for supervisors on employees' psychomental stress, absenteeism, and well-being. We included training aimed at improving supervisor-employee interaction, either off-the-job or on-the-job training, and training aimed at improving supervisors' capability of designing the work environment, either off-the-job or on-the-job training. In May 2019 we searched CENTRAL, MEDLINE, four other databases, and most relevant trials registers (ICTRP, TroPHI, ClinicalTrials.gov). We did not impose any language restrictions on the searches. We included randomised controlled trials (RCT), cluster-randomised controlled trials (cRCT), and controlled before-after studies (CBA) with at least two intervention and control sites, which examined the effects of supervisor training on psychomental stress, absenteeism, and well-being of employees within natural settings of organisations by means of validated measures. At least two authors independently screened abstracts and full texts, extracted data and assessed the risk of bias of included studies. We analysed study data from intervention and control groups with respect to different comparisons, outcomes, follow-up time, study designs, and intervention types. We pooled study results by use of standardised mean differences (SMD) with 95% confidence intervals when possible. We assessed the quality of evidence for each outcome using the GRADE approach. We included 25 studies of which 4 are awaiting assessment. The 21 studies that could be analysed were 1 RCT, 14 cRCTs and 6 CBAs with a total of at least 3479 employees in intervention and control groups. We judged 12 studies to have an unclear risk of bias and the remaining nine studies to have a high risk of bias. Sixteen studies focused on improving supervisor-employee interaction, whereas five studies aimed at improving the design of working environments by means of supervisor training.Training versus no interventionWe found very low-quality evidence that supervisor training does not reduce employees' stress levels (6 studies) or absenteeism (1 study) when compared to no intervention, regardless of intervention type or follow-up. We found inconsistent, very low-quality evidence that supervisor training aimed at employee interaction may (2 studies) or may not (7 studies) improve employees' well-being when compared to no intervention. Effects from two studies were not estimable due to missing data.Training versus placeboWe found moderate-quality evidence (2 studies) that supervisor training off the job aimed at employee interaction does not reduce employees' stress levels more than a placebo training at mid-term follow-up. We found low-quality evidence in one study that supervisor training on the job aimed at employee interaction does not reduce employees' absenteeism more than placebo training at long-term follow-up. Effects from one study were not estimable due to insufficient data.Training versus other trainingOne study compared the effects of supervisor training off the job aimed at employee interaction on employees' stress levels to training off the job aimed at working conditions at long-term follow-up but due to insufficient data, effects were not estimable. Based on a small and heterogeneous sample of controlled intervention studies and in contrast to prevailing consensus that supervisor behaviour influences employees' health and well-being, we found inconsistent evidence that supervisor training may or may not improve employees' well-being when compared to no intervention. For all other types of interventions and outcomes, there was no evidence of a considerable effect. However, due to the very low- to moderate-quality of the evidence base, clear conclusions are currently unwarranted. Well-designed studies are needed to clarify effects of supervisor training on employees' stress, absenteeism, and well-being.