The evidence for human tumor suppressor genes is reviewed. Initial evidence was provided by somatic cell hybridization, where somatic cell hybrids derived from the fusion of malignant and normal parental cells were found to be transformed but nontumorigenic. Tumorigenic segregants appeared at later intervals and were associated with the loss of specific normal chromosomes. Evidence for loss of tumor suppressor genes in many human malignancies was provided by a combination of cytogenetic and restriction fragment length polymorphism analyses. Functional analyses, using monochromosome transfer from normal cells into cancer cells, have confirmed the existence of suppressor genes and their critical role in control of tumor formation. Recently, the tumor suppressor gene Rb-1 has been cloned and also shown to have tumor-suppressing properties. Most recently, a candidate tumor suppressor gene on chromosome 17 (p53) has been implicated in colorectal carcinomas and other human malignancies. It is of interest to note that this gene was originally described as an oncogene. The biological mechanism of tumor suppression has been linked to the induction of differentiation in both somatic cell hybrids and osteosarcoma cells transfected with the normal Rb-1 gene. However, recent studies with monochromosome transfer into neuroblastoma cells indicates that differentiation may be dissociated from tumor suppression. Tumor suppressor genes do not act directly as negative regulators of conventional "dominantly-acting" oncogenes and therefore cannot be considered as anti-oncogenes in the sense of directly interacting with and regulating the expression of such oncogenes as ras and myc. However, it is speculated that they may negatively regulate an, as yet undiscovered, family of oncogenes which would not be dominantly expressed.