Lipophilic positively-charged compounds are facilitated across biological membranes by the transmembrane potential of intact cells. One such compound, rhodamine 123, has recently been shown to be selectively toxic toward a variety of transformed (carcinoma), epithelial cells in vitro (Lampidis et al., 1982; Bernal et al., 1982; Lampidis et al., 1983). A mechanism that could account for the selectivity of this agent would be a difference in the plasma membrane potential between normal and carcinoma cells. We report here that a significantly higher transmembrane potential has been found in a pair of carcinoma (83 mV for human breast and -99 mV for human cervix) as compared to normal (-56 mV for marsupial kidney and -48 mV for monkey kidney) epithelial cell lines. We also identified 3 other positively-charged lipophilic compounds, safranin 0, rhodamine 6G and tetraphenylphosphonium chloride (TPP+), which show selective toxicity toward carcinoma cells in vitro, while an uncharged lipophilic analog, rhodamine 116, does not. These data suggest that the higher plasma membrane potential of carcinoma cells may in part contribute to the preferential accumulation and selective toxicity of the lipophilic cationic compounds we have examined. An extension of this concept to an in vivo environment could lead to a class of cationic compounds which selectively exploit differences between normal and carcinoma cells.