A target is autonomically tested using a number of test cases. Dependencies among the test cases exist. A dependency between a first test case and a second test case means that where testing of the target using the first test case results in failure, testing of the target using the second test case is not performed. The dependencies among the test cases are constructed using one or more dependency rules. The dependency rules are updated while testing of the target using the test cases is performed, such that new dependencies among the test eases revealed during testing are incorporated into the dependency rules. Autonomic testing in this way provides for more intelligent testing of the target without human interaction and/or human intervention.