We discuss the problem of oncogenic hazard evaluation for new chemicals. In the recent past, assessment of global carcinogenicity in rodents was considered the most significant type of information. It has recently emerged that this type of information is not adequate to distinguish initiation and genotoxicity from promotion-like effects. In this report we suggest that hazards from initiating agents and hazards from promoting agents should be treated separately. In this perspective, longterm experiments for carcinogenicity in rodents should still play an important role but a less central one. Initiation-promotion experiments in different target organs are recommended. We suggest a strategy of hazard evaluation related to the initiating potential as distinct from overall carcinogenicity. The problem of utilizing not only the qualitative component of the available information, but also the quantitative component, is considered. Finally, we discuss a possible hazard evaluation for promoting effects, but conclude that this area is very much in its infancy. Possible contributions of computer science technology to the problem of oncogenic hazard evaluation are briefly introduced.