Many diabetic patients in general practice do not achieve good glycaemic control. The aim of this study was to assess which characteristics of type 2 diabetes patients treated in primary care predict poor glycaemic control (HbA1c > or = 7%). Data were collected from the medical records. 1641 patients were included who had mean HbA1c 7.1(SD 1.7)% , and 42% had HbA1c > or = 7%. On univariate analysis younger age; longer duration of diabetes; higher levels of blood glucose at diagnosis; most recent fasting blood glucose (FBG), total cholesterol, and triglyceride; higher body mass index (BMI); treatment with oral hypoglycaemic agents (OHA); treatment with insulin; more GP-visits for diabetes in the last year; and lower educational level were associated with poor control. Both in multiple linear regression and in multiple logistic regression higher levels of FBG (odds ratio (OR): = 1.6, 95% confidence interval (CI): 1.49, 1.70), treatment with OHA (OR: 2.1, 95% CI: 1.41, 3.04), treatment with insulin (OR: 7.2, 95% CI: 4.18, 12.52), lower educational level (OR: 1.26, 95% CI: 1.01, 1.56) were independently associated with poor levels of HbA1c. When FBG levels were excluded from the model, higher blood glucose at diagnosis, higher values for triglyceride and total cholesterol, and younger age predicted poor glycaemic control, but these variables explained only 15% of the variation in HbA1c. In conclusion prediction of poor glycaemic control from patient characteristics in diabetic patients in general practice is hardly possible. FBG appeared to be a strong predictor of HbA1c, which underlines the usefulness of this simple test in daily diabetes care. The worse metabolic control in those treated with either OHA or insulin suggests that current treatment regimes might be not sufficiently applied to reach the targets of care. Providers of diabetes care should be attentive to patients with lower educational level.