Due to the diffusion of the internet and the increase in the number of politicians who attack mass media and gain support, the problem of a decrease in the public’s trust in mass media is gaining attention both in Japan and abroad. However, the wording of questions measuring trust in mass media (TVs, newspapers, and magazines) is not consistent; hence it is difficult to examine whether trust is declining or not. We reviewed the previous findings of representative social surveys and revealed three differences in question wording: (1) degree adverbs assigned to choices of 4-point scales, (2) whether to measure trust in newspapers and magazines together, (3) whether to clearly indicate that the target of the question is an organization. We conducted a randomized web survey experiment to explore the differences in expressed trust based on question wording. As a result, level of the trust in media differs by up to 25％ or more depending on the question wording. The findings indicate the importance of choosing question wording with a clear reason in measuring trust in media.