A bystander effect is described when nontransduced or genetically unmodified cells are killed during death of genetically modified tumor cells transduced with a suicide gene. The "bystander effect" greatly enhances the efficacy of the herpes simplex virus-thymidine kinase/ganciclovir (HSV-TK/GCV) gene therapy approach for cancer. The mechanism of the bystander effect is controversial. In this study, we examined the role of intercellular gap junction communication (GJIC) for the bystander effect in human gastrointestinal tumor cells. Our results show that the extent of the bystander effect varied amongst the tumor cell lines; pancreatic cancer cells BXPC-3 exhibited excellent bystander effects in vitro and in vivo studies whereas other gastrointestinal tumor cell lines such as pancreatic cancer cells MIAPACA-2, and colon cancer cells HT-29 showed poor bystander effects. Bystander effects were only found in the presence of cell-to-cell contact. The extent of the bystander effect was independent of the level of HSV-TK activity in the transduced tumor cells and was correlated with GJIC as demonstrated by an in vitro dye-transfer assay. Expression of the mRNA levels of gap junction protein connexin 43 was 8- to 26-fold or greater and connexin 26 gene expression was 2- to 229-fold greater in BXPC-3 cells compared to HT-29, MIAPACA-2, and PANC3 cells. Our results suggest that intercellular communication is essential for the bystander effect. The correlation between GJIC and the extent of the bystander effect suggest a role for GJIC in mediating the bystander effect. Analysis of tumors for GJIC or expression of gap junction proteins may identify the subset of patients suitable for gene therapy with the HSV-TK/GCV approach.