<p>The evidence that mindfulness-based programmes (MBPs) improve the mental health and well-being of social workers is growing. Which MBP mechanisms improve these outcomes is unclear. This study attempts to identify what these mechanisms might be, using data from two RCTs investigating the impact of the MBP participation on social worker depression, anxiety, and mental well-being. Structural equation models (SEM) were constructed on the data of the 94 social workers who completed a MBP. Conditional direct, and indirect effect models of changes in mindfulness, self-compassion, attention regulation/decentering, acceptance, non-attachment, aversion, mediating variables (rumination and worry) and outcomes (depression, anxiety, and psychological well-being) reflecting the clinically modified Buddhist psychological model (CBPM) were tested. This study provides initial evidence that direct and mediated effects, and mediated effects CBPM models of social worker depression, anxiety and mental well-being might useful explanatory frameworks of how these outcomes improve through MBP participation. The pattern of significant direct and mediated effects found indicate that multiple outcomes could be improved through multiple mechanisms. Mindfulness, worry, and rumination appear to function as universal mechanisms of action of MBP participation on social worker depression, anxiety, and mental well-being. Attention regulation/decentering appears to function as a universal mechanism of action on depression and anxiety.</p>