Software testing is becoming more and more important as it is a widely used activity to ensure software quality. Testing is now an essential phase in software development life cycle. Test execution becomes an activity in the critical path of project development. In this case, a good test execution effort estimation approach can benefit both tester managers and software projects.This paper proposes an experience-based approach for test execution effort estimation. In the approach, we characterize a test suite as a 3-dimensions vector which combines test case number, test execution complexity and its tester together. Based on the test suite execution vector model, we set up an experience database, and then a machine learning algorithm is applied to estimate efforts for given test suite vectors. We evaluate the approach through an empirical study on projects from a financial software company.