Mindfulness meditation is a set of attention-based, regulatory and self-inquiry training regimes. Although the impact of mindfulness meditation training (MT) on self-regulation is well established, the neural mechanisms supporting such plasticity are poorly understood. MT is thought to act on attention through interoceptive salience and attentional control mechanisms, but until now conflicting evidence from behavioral and neural measures has made it difficult to distinguish the role of these mechanisms. To resolve this question we conducted a fully randomized 6-week longitudinal trial of MT, explicitly controlling for cognitive and treatment effects with an active control group. We measured behavioral metacognition and whole-brain Blood Oxygenation Level Dependent (BOLD) signals using functional MRI during an affective Stroop task before and after intervention. Although both groups improved significantly on a response-inhibition task, only the MT group showed reduced affective Stroop conflict. Moreover, the MT group displayed greater dorsolateral prefrontal cortex (DLPFC) responses during executive processing, consistent with increased recruitment of top-down mechanisms to resolve conflict. In contrast, we did not observe overall group by time interactions on negative affect-related RTs or BOLD responses. However, only participants with the greatest amount of MT practice showed improvements in response-inhibition and increased recruitment of dorsal anterior cingulate cortex (dACC), medial prefrontal cortex (mPFC), and right anterior insula during negative valence processing. Collectively our findings highlight the importance of active control in MT research, and indicate unique neural mechanisms for progressive stages of mindfulness training.