Continuous integration is an important but time and resource consuming process. Testing should be continuously done during this period. Developers need to select and refine test case as soon as possible to reduce testing cost. However, testing in continuous integration frequently is prohibitively expensive. To address this issue, we improve RETECS method using FP-Growth algorithm. The relationship of test case is identified by FP-Growth, a frequent pattern mining algorithm. Test cases are reprioritized using adjusting rule formed by FP-Growth algorithm. Our test case reprioritization is based on history testing data and prioritization by RETECS method. The laboratory results express that our approach can improve fault detection ratio compared to RETECS method.