OBJECTIVE - To determine the effect of a bicultural community health worker (CHW) on completion of diabetes education in an inner-city. Hispanic patient population and to evaluate the impact of completion of the education program on patient knowledge, self-care behaviors, and glycemic control. RESEARCH DESIGN AND METHODS - Patients were randomized into CHW intervention and non-CHW intervention groups. All patients received individualized, comprehensive diabetes education from a certified diabetes nurse educator after baseline demographic information, diabetes knowledge, diabetes self- care practices, and glycohemoglobin levels were assessed. Rates of education program completion were determined diabetes knowledge, self-care practices, and glycohemoglobin levels were reassessed at program completion and at a later postprogram follow-up medical appointment and compared to baseline. Logistic regression analysis and the Mantel-Haenszel χ2 statistic were used to determine the effect of the CHW assignment on program completion. Analyses of covariance were performed with end-of-treatment behavior scores, knowledge scores, and glycohemoglobin levels as outcome variables, controlling for baseline values and testing for the effect of CHW assignment. RESULTS - Of 64 patients enrolled in the study 40 (63%) completed and 24 (37%) dropped out before completing the diabetes education program. Of the patients having CHW intervention, 80% completed the education program, compared with 47% of patients without CHW intervention (P = 0.01). 'Dropouts' were younger (age 47.5 ± 12.5 years [mean ± SDI]) compared with patients who completed the program (55.9 ± 9.9 years) (P = 0.004). Dropout status showed no significant relationship to educational level achieved or literacy level. For the program 'completers,' knowledge levels and selected self-care practices significantly improved, and glycohemoglobin levels improved from a baseline level of 11.7% to 9.9% at program completion (p = 0.004) and 9.5% at the postprogram follow- up (P < 0.001). The effect of the CHW assignment on program completion, controlling for financial status and language spoken, was extremely robust (P = 0.007). The effect of the CHW on knowledge, self-care behavior, or glycohemoglobin outcome variables was not statistically significant. CONCLUSIONS - These findings suggest that intervention with a bicultural CHW improved rates of completion of a diabetes education program in an inner- city Hispanic patient population irrespective of literacy or educational levels attained. Our data further suggests that completion of individualized diabetes educational strategies leads to improved patient knowledge, self- care behaviors, and glycemic control.