In studies of risk assessment due to environmental exposures, we are interested in the mortality due to a particular disease when several other diseases are acting on a person's life. The method of analyzing survival data in such a population is called “competing risk analysis.” We also have a fundamental assumption that each death takes place due to a single cause. When an individual is put on test, only his time of death and cause of death are recorded. This competing risk data provides insufficient information to determine the joint distribution of the survival times giving rise to an identifiability problem. In this article, the cause‐specific hazard functions are introduced and their estimation and testing, for modeling, both independent and dependent lifetimes are discussed. The identifiability problem is also addressed and some ways to deal with the competing risk data in the presence of association between the latent survival times are suggested.