In a car accident which is involving pedestrians, head injuries occur very frequently as the head of the pedestrian hits the windshield. The head injury criterion (HIC) obtained through the windshield impact test is used to evaluate the pedestrian injury, and car manufacturers are trying to meet the criterion by changing the design and/or materials.. However, there are some difficulties in the windshield impact test, e.g. a large scatter of the test data or windshield shape-dependent property of the test. These problems make it very difficult to obtain the meaningful results from single test and thus, tests should be executed several times. In this study, a lab-scale windshield impact test was performed using a modified instrumented dart impact (IDI) tester. Tests were carried out by switching test conditions such as the impact speed, the size of the head form and the specimen thickness. The key results such as acceleration and displacement curves of a head form, peak values from the acceleration curve, cracking modes, etc. were compared for various test conditions. In addition, the numerical simulation was carried out to correlate the lab-scale test results with full-scale windshield impact test, and the correlation between the lab-scale simulation result and full-scale simulation result of the earlier study is discussed.