Objective: Despite the frequent use of insulin in diabetes patients, it is unclear exactly how much blood glucose levels are decreased by a unit of insulin. This study was aimed to evaluate the exact glucose-lowering effect of insulin. Methods: Retrospective analysis from 2008 to 2018 on diabetes patients treated with multiple daily insulin injections (MDI) within Seoul National University Hospital was performed. Glucose levels were analyzed by dividing it into fasting and postprandial glucose. Change of oral anti-diabetic drugs or use insulin other than basal and rapid-acting insulins during treatments were excluded. When the glucose-lowering effects on both fasting and postprandial glucose were better than the median, those patients were defined as the good responder. The logistic regression model was used to evaluate the risk factors for the response. Results: Among 845 adult patients with diabetes (4135 days of MDI), mean age was 58 ± 16, and 150 (17.8%) had type 1 diabetes, and total daily insulin dose was 50.7 ± 24.4 U (basal insulin, 23.2 ± 10.7 U). Per unit of basal insulin, 7.5 ± 18.7 mg/dL of fasting glucose level was decreased, and postprandial glucose level was decreased by 15.5 ± 26 mg/dL per unit of rapid-acting insulin. Good responses were associated with female (Odd ratio [OR] 1.91, 95% CI 1.23-2.97, P = 0.004) and lower BMI (OR 0.94, 95% CI 0.88, 0.99, p-value=0.025). Patients with type 1 diabetes showed a tendency for good response (OR 1.67, 95% CI 0.93-3.00, P = 0.088). Conclusion: In this real-world analysis, actual glucose-lowering effects of insulin were less than those previously know. Female and lower BMI were associated with good response.