People with diabetes frequently have elevated diabetes distress. Although mindfulness-based stress reduction (MBSR) therapy has been shown effective in reducing diabetes distress, it has only been delivered by psychologists or a multidisciplinary team with an attrition rate of up to 39%, which limits its dissemination to a broader audience. This study was aimed to pilot evaluate the feasibility of a nurse-led MBSR therapy and explore its potential efficacy amongst people with type 2 diabetes mellitus. A total of 100 participants were randomly allocated either to the intervention group (nurse-led MBSR therapy + regular diabetes education) or the control group (regular diabetes education). Data on diabetes distress, diabetes self-efficacy, and diabetes self-management were collected at baseline, 8 and 12 weeks. Hemoglobin A1c (HbA1c) was collected at baseline and 12 weeks. A generalized estimating equation analysis for repeated measures was used to determine intervention and time effects. As predicted, the nurse-led MBSR therapy had a significant time-by-group interaction effect on diabetes distress total score (95% confidence interval [CI]: 0.58–0.77, p < 0.001), diabetes self-efficacy (95% CI: −0.93 to −0.74, p < 0.001), diabetes self-management (95% CI: −10.80 to −7.83, p < 0.001), and HbA1c levels (95% CI: 0.04–1.14, p = 0.03) in the intervention group compared with the control group over 12 weeks. This is the first nurse-led MBSR therapy in a hospital setting that is feasible and has the potential to improve health outcomes. This approach may offer an innovative model to deliver MBSR therapy. A randomized controlled trial comparing the nurse-led MBSR plus usual diabetes education with usual diabetes education along with a mechanism to equalize intervention attention between the groups is indicated.