Virus-mediated oncolysis is a rapidly growing field with the potential to dramatically alter the future of cancer therapy. Replication-selective viruses are superior to non-replicating vectors in several aspects, such as the amplification of the initial low-dose viral inoculum up to 10(3)-10(3)-fold, lateralization into neighboring cells, introduction of novel cell killing mechanisms, and a potential for a safe profile. However, due to their capacity to replicate, the importance of tumor selectivity is further underscored. Of the replication-selective viruses, adenoviruses (Ad) possess several attributes that appear essential for targeting and eliminating tumor cells. These include susceptibility to genomic modifications, convergence with cellular pathways implicated in carcinogenesis, and a high oncolytic capacity. A primary tumor targeting strategy of oncolytic Ad is based on re-engineering the viral genome viruses to construct conditionally replicative adenoviruses (CRAds). In this regard, modification of CRAd genome is traditionally designated as type I or type II. Type I CRAds are based on mutation or deletion of early Ad genes. Type II CRAds are based on the placement of essential early Ad genes under tissue/tumor-specific regulatory elements in a heterologous context. Thus, both strategies confer varying degrees of tumor-specific replication. Recent data, however, indicate that type III CRAds, embodying the paradigms of both type I and II, offer better replication selectivity for tumor cells while maintaining efficient oncolysis. These characteristics of CRAds yield therapeutic indices unprecedented heretofore in cancer therapy. However, other biological aspects of CRAds should also be addressed before these agents prove as first-line antitumor agents. When these issues are resolved, novel tumor cell killing potential of CRAds may truly be realized and dramatically alter future cancer therapy.