PPROXIMATELY A DECADE AGO, THE MOLECULAR GENETIC era of human cancer was ushered in with the discovery of" dominantly acting" activated cellular oncogenes (1). The first activated oncogenes were isolated by transfection of DNA from human cancer cells into mouse NIH 3T3 cells, a process that resulted in neoplastic transformation. The activated oncogenes were quickly found to be homologs of retroviral transforming genes (2). This finding, which was predicted on the basis ofthe seminal studies that showed that the avian retroviral src oncogene had evolved from the capture of a cellular protooncogene (3), led to the further identification of numerous candidate cellular oncogenes. The discovery that activated oncogenes could be found in 10 to 30 percent of human cancers led to theories that activation of single or multiple cooperating cellular oncogenes was in itself sufficient to create a cancerous cell. These theories were all the more attractive when it was found that the expanding list of oncogene functions included growth factors, growth factor receptors, signal transducers, protein kinases, and transcriptional activators-all of which, when behaving aberrantly, might lead to uncontrolled cell proliferation. A tacit assumption in many interpretations of these studies was the dominant nature of activated oncogenes. However, earlier studies with somatic cell hybrids had clearly shown that when malignant cells were fused with normal cells, the resulting hybrid cells were nontumorigenic (4). This phenomenon oftumor suppres-sion indicated that a gene (or genes) from a normal cell might replace a defective function in the cancer cell and render it respon-sive to normal regulators of cell growth. The notion of loss of genetic function being a critical event in the genesis of cancer received further support when it was shown by a combination of cytogenetic and molecular studies of restriction fragment length polymorphisms (RFLP's) that specific chromosomal deletions are often associated with certain human malignancies (5). The combination ofthese studies led to the hypothesis that a class of genetic elements, termed tumor suppressor genes, exist which must be inactivated in some fashion-for example, by deletion, point mutation, or methylation-before a cell can become malignant (6).