This study investigates the effect of audio sources that Korean teachers utilize when developing listening test items. Before developing listening test items, teachers consider learners’ competency and learning targets to set the goal of the test. And after confirming if topics and contents are suitable for the test, teachers collect materials and try to make a meaningful listening test utilizing audio sources as much as possible. However, issues arise at this point: Most of the contents teachers use for a listening test are from written media. Teachers then change those written contents into spoken ones, such as dialogues or monologues. Listening test items derived from these contents are in fact not different from reading test items, and it causes reading questions appearing in a listening test. Assuming this kind of test item production somewhat restricts proper evaluation for learners’ listening ability, the researcher of this paper looked into how materials used in making listening test affected the test items. In her experiment, well-experienced Korean teachers were divided into two groups: One group was given materials from audio resources, and the other group was given materials from written one. The two groups were then asked to produce listening test items. In the result, the latter group was turned out to have developed better test items which are more closed to the nature of listening.