Nucleoside analogues which show antimetabolic activity in cells have been successfully used in the treatment of various tumors. Nucleosides such as 1-beta-D-arabinofuranosylcytosine (araC), 6-mercaptopurine, fludarabine and cladribine play an important role in the treatment of leukemias, while gemcitabine, 5-fluorouracil and its prodrugs are used extensively in the treatment of many types of solid tumors. All of these compounds are metabolized similarly to endogenous nucleosides and nucleotides. Active metabolites interfere with the de novo synthesis of nucleosides and nucleotides or inhibit the DNA chain elongation after being incorporated into the DNA strand as terminators. Furthermore, nucleoside antimetabolites incorporated into the DNA strand induce strand-breaks and finally cause apoptosis. Nucleoside antimetabolites target one or more specific enzyme(s). The mode of inhibitory action on the target enzyme is not always similar even among nucleoside antimetabolites which have the same nucleoside base, such as araC and gemcitabine. Although both nucleosides are phosphorylated by deoxycytidine kinase and are also good substrates of cytidine deaminase, only gemcitabine shows antitumor activity against solid tumors. This suggests that differences in the pharmacological activity of these nucleoside antimetabolites may reflect different modes of action on target molecules. The design, in vitro cytotoxicity, in vivo antitumor activity, metabolism and mechanism of action of sugar-modified cytosine nucleosides, such as (2'S)-2'-deoxy-2'-C-methylcytidine (SMDC), 1-(2-deoxy-2-methylene-beta-D-erythro-pentofuranosyl)cytosine (DMDC), 1-(2-C-cyano-2-deoxy-1-beta-D-arabino-pentofuranosyl)cytosine (CNDAC) and 1-(3-C-ethynyl-beta-D-ribo-pentofura-nosyl)cytosine (ECyd), developed by our groups, are discussed here.