The operation of a large-scale metro system creates problematic interior noise; the impact of this noise on passengers and drivers is a subject of increasing concern. To investigate the quantitative relationship between metro interior noise and passengers' annoyance, this study analyzed questionnaires on passenger annoyance completed by 118 volunteers. The feedback from the questionnaire concerned eleven metro lines in Beijing. To test the interior noise levels, the volunteers were divided into two groups: A and B. The volunteers in group A took the same metro train as the testers, whereas those in group B took different trains. A total of 2080 noise annoyance samples from metro tunnel sections were collected and analyzed. Finally, the exposure-response relationship between interior noise and passenger annoyance was obtained by fitting these data with a logistic function. The results indicated that there was a significant positive correlation between the average subjective annoyance and the averaged equivalent sound pressure level. The fitting result was better for group A than for group B. For the mixed samples of two groups, the fitting result was greatly affected by the contribution of group A. To provide an acoustically comfortable environment, metro interior noise should not exceed 84-85 dB(A).