Cancer drugs such as 5-fluorouracil (5-FU) that target the enzyme thymidylate synthase (TS) have been and are still being widely used in cancer treatment, but as with other anti-cancer drugs, the majority of tumors do not respond to the treatment, whereas the patients still suffer drug-related toxicity. The most recent attempts at improving cancer treatment have taken the pharmacogenetic approach of identifying biochemical response determinants for response, so that patients with suboptimal determinants who unlikely to respond can be identified prior to treatment. Studies to date indicate that high intratumoral levels of TS gene expression or TS protein generally predict for non-response, whereas low levels are associated with a high response rate. Measuring these determinants requires tumor tissue and, in the case of gene expression, a technically demanding quantitative PCR procedure. Thus, considerable interest was generated by data suggesting that the variable number of a 28 base-pair (bp) segment in the promoter region of the TS gene was associated with TS gene expression and/or protein expression, as well as with tumor response to 5-FU therapy, toxicity and patient survival. However, not all studies have obtained the same results, so that the role of this TS polymorphism as a predictor of treatment outcome is still not clear and is currently under evaluation. This review will summarize pharmacogenomic studies of TS that were aimed at elucidating the function of this genetic polymorphism.