Developers must select a high-performance fault localization (FL) technique from available ones. A conventional approach is to try to select only one FL technique that is expected to attain high performance before debugging activity. In contrast, we propose a new approach that dynamically selects better FL techniques during debugging activity.