Diabetes is challenging because of its chronicity and high demands for daily monitoring and care. Diabetic patients must modify their daily lives to comply with diet, exercise, medication taking and glucose testing. Although numerous clinical studies have emphasized the importance of behavioral and emotional influences on glycemic control among diabetic patients, relatively few studies have reported about noninsulin-dependent (NIDDM) adult diabetic patients in longitudinal studies. The present study was designed to predict change in glycemic control during 6 months based on demographic and psychosocial factors among middle-aged and older patients with NIDDM. Subjects were 61 outpatients with NIDDM who completed both the initial and the 6-month follow-up assessments (out of 74 at the beginning) at the Asahi Life Foundation Institute for diabetic care and research. Forty-five were male, sixteen were female and subjects' ages ranged from 36 to 80 years, with a mean of 58.0 years. Patients were requested to fill out diabetic checklist and several psychological questionnaires, from which 9 demographic and 10 psychosocial variables were selected as predictor variables. Level of glycemic control was assessed by glycosylated hemoglobin (HbA_1c) assays at the beginning of the study (pre-HbA_1c) and at the 6-month follow-up (post-HbA_1c) . '% change of HbA_1c' = { (post-HbA_1,c-pre-HbA_1c) /pre-HbA_1c} x 100 was then used as a criterion variable. Multivariate statistical analysis, the Quantification type I method, was conducted in order to take account of potential influence between each of predictor variables and to utilize categorical types of predictor variables. Mean % change of HbA_1c was 0.20% (range : -22.0% to 33.3%) . It was revealed that the combination of egogram pattern, treatment regimen and diabetic education could most accurately predict % change of HbA_1c (R-squared=0.344. F= 5.76,p=0.0002) . The egogram pattern based on transactional analysis which indicating the ways of interpersonal relationships was the best predictable factor. It was indicated that patients with NP dominant pattern had deteriorated the most and those with reverse 'N' pattern had improved the most. It is concluded that useful information to predict % change of level of glycemic control may be obtained from the longitudinal study using the combination of certain psychosocial factors. Future research should refine the measurement to evaluate diabetic self-care behavior, and increase the number of subjects from other institutes.