Gap junctional intercellular communication (GJIC) is considered to play a key role in the maintenance of tissue independence and homeostasis in multicellular organisms by controlling the growth of GJIC-connected cells. Gap junction channels are composed of connexin molecules and, so far, more than a dozen different connexin genes have been shown to be expressed in mammals. Reflecting the importance of GJIC in various physiological functions, deletion of different connexin genes from mice results in various disorders, including cancers, heart malformation or conduction abnormality, cataract, etc. The possible involvement of aberrant GJIC in abnormal cell growth and carcinogenesis has long been postulated and recent studies in our own and other laboratories have confirmed that expression and function of connexin genes play an important role in cell growth control. Thus, almost all malignant cells show altered homologous and/or heterologous GJIC and are often associated with aberrant expression or localization of connexins. Aberrant localization of connexins in some tumour cells is associated with lack of function of cell adhesion molecules, suggesting the importance of cell-cell recognition for GJIC. Transfection of connexin genes into tumorigenic cells restores normal cell growth, supporting the idea that connexins form a family of tumour-suppressor genes. Some studies also show that specific connexins may be necessary to control growth of specific cell types. We have produced various dominant-negative mutants of Cx26, Cx32 and Cx43 and showed that some of them prevent the growth control exerted by the corresponding wild-type genes. However, we have found that connexins 32, 37 and 43 genes are rarely mutated in tumours. In some of these studies, we noted that connexin expression per se, rather than GJIC level, is more closely related to growth control, suggesting that connexins may have a GJIC-independent function. We have recently created a transgenic mouse strain in which a mutant Cx32 is specifically overexpressed in the liver. Studies with such mice indicate that Cx32 plays a key role in liver regeneration after partial hepatectomy. A decade ago, we proposed a method to enhance killing of cancer cells by diffusion of therapeutic agents through GJIC. Recently, we and others have shown that GJIC is responsible for the bystander effect seen in HSV-tk/ganciclovir gene therapy. Thus, connexin genes can exert dual effects in tumour control: tumour suppression and a bystander effect for cancer therapy.