Mindfulness-based programs (MBPs) are increasingly offered in the workplace to reduce stress and enhance wellbeing. A growing body of evidence suggests that these programmes can be beneficial; however, dropout rates are often high, and there is great heterogeneity in both programme content and strength of effects between studies. This thesis contributes to a better understanding of what makes these programmes successful, under what circumstances, for whom, and why. A realist approach was employed that is theory-driven and well-suited to make sense of complex interventions, such as MBPs. First, I conducted a realist review where I analysed and synthesised data from 75 existing studies of workplace MBPs to identify the mechanisms through which these programmes work or do not work and discover how particular contexts impact these mechanisms and, consequently, outcomes. The findings were consolidated into a programme theory augmented by theories from organisational literature, such as conservation of resources (COR) theory and psychological safety. I then tested and refined parts of that programme theory with realist evaluation methodology. The evaluation consisted of five workplace MBPs and included quantitative and qualitative data collection methods. The research results show that for managers and employees to invest in an MBP and benefit from it, it is crucial that they feel safe to engage with selfcare at work and show vulnerability to each other. The more participants feel that they are not at work, the safer they feel. To be beneficial, it is also important that managers and employees link the programme and its activities to existing goals and practices. If these conditions are not met, the costs of investing in an MBP may be too high and lead to the strategic use of brief mindfulness exercises, non-adherence, or dropout. These findings should be of value to decision makers and practitioners wishing to offer and implement effective and sustainable workplace MBPs.