This meta-analysis was conducted to assess the effect of educational intervention on the glycemic control of type 2 diabetes. We searched for English language trials in the Medline database (1986-2005). Studies were included if they were randomized controlled trials, tested the effect of health education on adult patients, and reported extractable data on the effect of treatment on glycohemoglobin (GHb). A total of 30 studies of 115 initially identified articles met these selection criteria. We classified the studies into one of the following categories by the type of intervention: diet and exercise, use of education tools, educational approach, mode of instruction, and telephone follow-up. We then calculated the pooled effect sizes by using random effect models. The largest decrease in GHb was observed with diet and exercise intervention (standardized difference in means (SDM)=-0.722, 95% confidence interval (CI)=-1.032--0.412). It was difficult to reduce the body weight and GHb by diet management alone, so exercise needed to be incorporated into the program. Since telephone follow-up was also effective (SDM=-0.229, 95% CI=-0.444--0.014), using a telephone instead of visits would be a feasible option for planning sustainable education programs for workers.