Improvements in stable, or dispositional, mindfulness are often assumed to accrue from mindfulness training and to account for many of its beneficial effects. However, research examining these assumptions has produced mixed findings, and the relation between dispositional mindfulness and mindfulness training is actively debated. A comprehensive meta-analysis was conducted on randomized controlled trials (RCTs) of mindfulness training published from 2003-2014 to investigate whether (a) different self-reported mindfulness scale dimensions change as a result of mindfulness training, (b) key aspects of study design (e.g., control condition type, population type, and intervention type) moderate training-related changes in dispositional mindfulness scale dimensions, and (c) changes in mindfulness scale dimensions are associated with beneficial changes in mental health outcomes. Scales from widely used dispositional mindfulness measures were combined into 5 categories for analysis: Attention, Description, Nonjudgment, Nonreactivity, and Observation. A total of 88 studies (n = 5,787) were included. Changes in scale dimensions of mindfulness from pre to post mindfulness training produced mean difference effect sizes ranging from small to moderate (g = 0.28-0.49). Consistent with the theorized role of improvements in mindfulness in training outcomes, changes in dispositional mindfulness scale dimensions were moderately correlated with beneficial intervention outcomes (r = .27-0.30), except for the Observation dimension (r = .16). Overall, moderation analyses revealed inconsistent results, and limitations of moderator analyses suggest important directions for future research. We discuss how the findings can inform the next generation of mindfulness assessment. (PsycINFO Database Record