Hospitals are a closed system and medial service failures are reluctant to be disclosed. Thus, few studies comprehensively explored medical service failure. Quality Control Circle (QCC) dataset of hospitals is one of the sources of service failure since QCC describe factors of service failure in detail. This study aims to shed light on the sensitive issue of hospital service failure through comprehensively analysing medical QCC data. This study uses a two-dimensional analytical framework constructed by service failure type (human error/ non-human error) versus problem type (problem-solving/ task achieving). A total of 2,080 entries of QCC data were collected. An innovative analysis method, integrating text analysis with content analysis method, is adopted. The results show the factors of service failure and their occurring frequencies. Moreover, the evolution scheme of the factors of service failure is provided. Finally, this study finds that there exist interaction effects among the service failure factors.