To assess whether self-monitoring of quantitative urine glucose or blood glucose is effective, convenient and safe for glycaemic control in non-insulin treated type 2 diabetes.Adults with non-insulin treated type 2 diabetes were recruited and randomized into three groups: Group A, self-monitoring with a quantitative urine glucose meter (n=38); Group B, selfmonitoring with a blood glucose meter (n=35); Group C, the control group without selfmonitoring (n=35). All patients were followed up for six months, during which identical diabetes care was provided.There was a significant decrease in HbA1c within each group (p<0.05). At the study conclusion, mean changes in HbA1c from baseline were −1.9% for Group A, −1.5% for Group B and −1.0% for Group C, and the proportion of patients achieving HbA1c≤6.5% were 38.9%, 35.3% and 20.0% respectively. However, no significant differences between the groups were found. The average monitoring frequency was significantly higher in Group A than in Group B. The incidence of hypoglycaemia and quality of life scores were similar between the groups.This study suggests that self-monitoring of urine glucose has comparable efficacy on glycaemic control, and facilitates better compliance than blood self monitoring, without influencing the quality of life or risk of hypoglycaemia.