WHAT IS KNOWN AND OBJECTIVE: Many drugs fail during development. However, detailed reasons for failure during drug development are almost never disclosed. We focused on the drugs whose clinical development and registration were initially hampered, but which were finally approved to identify reasons that delayed their marketing approval in Japan. METHODS: We analysed 727 new drug applications (NDAs) approved in Japan between 2001 and 2011. RESULTS AND DISCUSSION: Fifty-three NDAs had serious and identifiable problems during drug development. Of these, 43 NDAs had 'problem related to clinical data'. We found that the problems for withdrawal of these NDAs could be ascribed largely to inappropriate clinical data package and study design for supporting the intended indications and usage and to unclear clinical results for defining dosage regimen or efficacy of the drugs. WHAT IS NEW AND CONCLUSION: Our results indicate the importance of careful determination of the optimal dosage regimen and the choice of objective endpoints in clinical trials. Further, it is important to establish a clear strategy for generating the clinical data package, to include careful design of clinical trials on the basis of the nature of the target disease and target population. For drugs marketed in Japan, there is a need to include sufficient numbers of Japanese patients in the trials.