This proposal aims to undertake research on statistical problems arising in carcinogenicity experiments. The research program falls into four general categories: (1) analysis of carcinogenicity experiments when risks from intercurrent death do not act independently of the process tumor development; (2) the use of isotonic regression techniques in carcinogenicity testing and low-dose extrapolation; (3) the assessment of and adjustment for litter effects in time-to-tumor analyses; (4) exploratory analyses of the ED01 data base. The long range objectives of the proposal are to permit more efficient and sensitive analyses of carcinogenicity experiments through (i) a better understanding of the process of tumor development and subsequent death from tumor or other causes; (ii) a better understanding of the statistical methods used in carcinogenicity experiments, particularly when assumptions concerning tumor lethality, independence of risks, and classification of cause of death are violated and (iii) the development of statistical methods for situations where existing techniques are either inappropiate or make inefficient use of the data.