Cytotoxic nucleosides have proven to be ineffective for the treatment of hepatocellular carcinoma (HCC) due, in part, to their inadequate conversion to their active nucleoside triphosphates (NTP) in the liver tumor and high conversion in other tissues. These characteristics lead to poor efficacy, high toxicity, and a drug class associated with an unacceptable therapeutic index. Cyclic 1-aryl-1,3-propanyl phosphate prodrugs selectively release the monophosphate of a nucleoside (NMP) into CYP3A4-expressing cells, such as hepatocytes, while leaving the prodrug intact in plasma and extrahepatic tissues. This prodrug strategy was applied to the monophosphate of the well-known cytotoxic nucleoside cytosine-1-beta-D-arabinofuranoside (cytarabine, araC). Compound 19S (MB07133), in mice, achieves good liver targeting compared to araC, generating >19-fold higher cytarabine triphosphate (araCTP) levels in the liver than levels of araC in the plasma and >12-fold higher araCTP levels in the liver than in the bone marrow, representing a >120-fold and >28-fold improvement, respectively, over araC administration.