A hand-arm vibration syndrome, local vibration illness, occurs in some workers who use hand held vibration tools. It consists of white fingers, diffusely distributed finger neuropathy, pain in the hand and arm, and a small excess risk of osteoarthritis. This study is aimed to identify effective methods to confirm local vibration illness among various health exams, which are mentioned in worker's special health exam regulation. In addition, this study is aimed to quantitatively assess the daily vibration exposure level as a major determinant of vibration illness. The subjects, 46 vibration workers, were selected according to the results of the first special health exam about vibration hazards at shipbuilding industry in 1997. They all had experiences of work related blanching of fingers. Fifteen controls, who had no vibration exposure at all, were also recruited to compare their test results with the results of vibration workers. We adopted 1 subjective and 6 objective tests to evaluate the effectiveness and feasibility for confirming local vibration illness. These tests were history taking of subjective symptoms according to the Stockholm classification, checking blood pressure of finger, checking grasp power, checking finger skin temperature, nail-bed pressing test, vibration perceptional threshold test, and skin prick test for pain perception. Among these, checking skin temperature, nail-bed pressing test, and vibration perception test included cold water provocations. We also estimated some vibration exposure levels of hand held vibration tools by using previously published data from one automobile company. In conclusion, history taking of subjective symptoms according to the Stockholm classification, nail-bed pressing test, and vibration perceptional threshold test were discovered to be effective to diagnose local vibration illness. Furthermore, vibration perceptional threshold on right fingers showed a dose-response relationship to daily vibration exposure levels. The parameter beta was 0.0005(+/-0.0002), and statistically significant by REM (random effects model).