OBJECTIVE:Though cost data is essential for pharmacoeconomic study, the method of cost data collection is not standardized in Japan. We started cost data collection along with a clinical trial on breast cancer treatment(National Surgical Adjuvant Study of Breast Cancer: N-SAS BC02). The trial has been supported by Comprehensive Support Project for Oncological Research of Breast Cancer. The aim of this study is to demonstrate our approach for both direct and indirect cost data collection and discuss some issues raised through the process. METHODS: N-SAS BC02 is a randomized controlled trial of two regimens for breast cancer. From a societal perspective, both direct and indirect cost has been collected from a sample of 100 participants of the trial upon consent. For direct medical cost, we collected health insurance claims which were issued every month from each hospital for 1 year. Even though unit price of each medical procedure and drug is determined by the government,total cost of treatment is unsure, mainly because combination of procedures are different among health care providers. Direct non-medical cost includes cost for transportation and supportive devices. Indirect cost was defined as labor loss for participants. Direct non-medical cost and labor loss days were asked through a self-administered questionnaire from each participant. RESULTS: Electric data which were kept in a hospital for making health insurance claims were not easy to use because data format was not standardized. We propose to use standardized health insurance claim format for electronic submission. CONCLUSIONS:Though electronic submission is not common yet, it will be the best way to collect paper-based claims and put the electronic code for each procedure and drug for a while. Because it is difficult to collect those data retrospectively,it will be important to consider cost data collection especially direct non-medical and indirect cost along with a clinical trial.