Few studies have demonstrated an effect of educational interventions on glycaemic control in persons with Type 2 diabetes longer than 3-6 months after baseline. We aimed to investigate the effectiveness of an experience-based group educational programme 24 months after baseline and to pinpoint mediators that might play a role in achieving desired metabolic outcomes. We conducted a randomised controlled trial inviting self-referred persons with Type 2 diabetes (N=77 randomised). The pharmacist-led, year-long intervention was based on participants' experiences of glucose regulation during the monthly group discussions. We measured HbA1c at 0, 6, 12, and 24 months and a questionnaire was administered at baseline and final follow-up. Our findings indicated that participating in the intervention programme significantly decreased HbA1c by 0.4% at 24 months after baseline. Initial HbA1c, satisfaction with own diabetes-related knowledge, and treatment were found directly related to glycaemic outcomes. The intervention group exercised more in order to lower blood-glucose levels and was also more able to predict current blood-glucose levels before measuring it. Experience-based group education was effective in decreasing participants' HbA1c 1-year after completed intervention. Early effect of the intervention was followed by relapse after 12 months and a new, significant decrease at 24 months; this dual course implies that follow-up of educational interventions should involve several consecutive measurements to capture possible late effects. Both biomedical and subjective factors played a role in accounting for the variance of HbA1c at 2-year follow-up after baseline.