Self-compassion has been proposed as a mechanism of change in mindfulness-based programmes (MBPs). The current study systematically reviewed the evidence for the effect of MBPs on self-compassion, in randomised controlled trials addressing broad mental health outcomes (depression, anxiety and stress) in nonclinical populations, and statistically synthesisesd these findings in a meta-analysis.

Three databases were systematically searched, and pre-post programme between group effect sizes (Hedges g) were calculated and synthesised using meta-analytic procedures. Correlation between change in self-compassion and distress (r) was also assessed. Moderator analyses were conducted and publication bias was assessed.

Twenty-six studies met inclusion criteria (n = 598). A significant medium effect of pre-post change on self-compassion was found for MBPs compared to control conditions (g = 0.60, 95% CI = 0.41 to 0.80, p < 0.001). There was significant heterogeneity in the study sample, and no differences found for any of the moderators tested. There was no strong evidence for publication bias. Meta-analysis of correlation between change in self-compassion and distress was underpowered and found no significant effect. The improvement in self-compassion following MBI was not always consistent with improvements in depression or anxiety.

The results suggest that MBPs can increase self-compassion in nonclinical populations, though the moderators of this effect remain unknown. Methodological limitations include small sample sizes, over-reliance on wait-list control conditions and limitations in how self-compassion is measured. Theoretical and clinical implications of the review, and future research directions, are also discussed.