In an approach to cleanup of unpredictable test results, one or more computer processors generate a data area associated with a first test instruction in a test stream. The one or more computer processors determine whether the generated data area overlaps with an unpredictable data area. In response to determining the generated data area overlaps with an unpredictable data area, the one or more computer processors determine a second test instruction associated with the overlapped unpredictable data area, where the second test instruction precedes the first test instruction in the test stream. The one or more computer processors select a location in the test stream between the first test instruction and the second test instruction. The one or more computer processors insert one or more pre-requisite instructions in the selected location, where the one or more pre-requisite instructions load the overlapped unpredictable data area with pre-defined data.