A series of trichlorinated indole nucleosides has been synthesized and tested for activity against human cytomegalovirus (HCMV) and herpes simplex virus type-1 (HSV-1) and for cytotoxicity. Modifications of the previously reported 2,5,6-trichloro-1-(beta-d-ribofuranosyl)indole at the 3-position of the heterocycle were designed in part to test our hypothesis that hydrogen bonding is required at that position for antiviral activity. Analogues were synthesized using electrophilic addition at the 3-position or by synthesis of modified indole heterocycles followed by glycosylation and modification of the sugar. Among the modifications at the 3-position, only those analogues with hydrogen-bond-accepting character were active against HCMV (e.g., 3-formyl-2,5,6-trichloro-1-(beta-D-ribofuranosyl)indole, FTCRI, IC50 = 0.23 microM). Conversely, analogues with non-hydrogen-bonding substituents at the 3-position (e.g., 3-methyl-2,5,6-trichloro-1-(beta-D-ribofuranosyl)indole) were much less active (IC50 = 32 microM) than those with the requisite hydrogen-bonding capacity. The 5'-O-acyl analogue of FTCRI was obtained as an intermediate and also found to be a potent inhibitor of HCMV (IC50 < 0.1 microM). The synthesis of some additional 5'-O-acylated analogues did not provide a compound with increased antiviral activity. None of the indole nucleosides had significant activity against HSV-1, and none were cytotoxic to uninfected cells in their antiviral dose range. Results obtained from the antiviral evaluations have validated our hypothesis that hydrogen bonding at the 3-position is required for antiviral activity in this series of chlorinated indole nucleosides.