4'-thio-beta-D-arabinofuranosylcytosine (T-araC) exhibits excellent in vivo antitumor activity against a variety of solid tumors despite its structural similarity to beta-D-arabinofuranosylcytosine (araC), an agent which is poorly active against solid tumors in vivo. It is of great interest to elucidate why these compounds show a profound difference in antitumor activity. Because deoxycytidine kinase (dCK) is the critical enzyme in the activation of both compounds, here we report the differences in the substrate characteristics with human dCK between these compounds. The catalytic efficiency (V(max)/K(m)) of araC was 100-fold higher than that of T-araC using either ATP or UTP as the phosphate donor. However, V(max) values of araC and T-araC were similar when UTP was the phosphate donor. Since UTP is believed to be the true phosphate donor for dCK in intact cells, these data indicated that the rates of phosphorylation of these two compounds at high pharmacologically relevant concentrations would be similar. This prediction was confirmed in intact cell experiments, which supported the hypothesis that UTP is the physiological phosphate donor for dCK phosphorylation in cells. The relative lack of importance of phosphate donor to the phosphorylation of T-araC by dCK revealed important insights into the activation of this compound in human cells at pharmacological doses. These studies indicated that replacement of the 4'-oxygen with sulfur significantly reduced the substrate activity of nucleoside analogs with dCK and that the superior activity of T-araC with respect to araC against solid tumors was not due to superior activity with dCK.