We have previously shown that a series of nonnucleoside pyrrolo[2,3-d]pyrimidines selectively inhibit the replication of herpes simplex virus type 1 (HSV-1) and human cytomegalovirus (HCMV). These compounds act at the immediate-early or early stage of HCMV replication and have antiviral properties somewhat similar to those of roscovitine and olomoucine, specific inhibitors of cyclin-dependent kinases (cdks). In the present study we examine the hypothesis that pyrrolo[2,3-d]pyrimidines exert their antiviral effects by inhibition of cellular cdks. Much higher concentrations of a panel of pyrrolo[2,3-d]pyrimidine nucleoside analogs with antiviral activity were required to inhibit recombinant cdk1/cyclin B compared to the submicromolar concentrations required to inhibit HCMV and HSV-1 replication. 4,6-Diamino-5-cyano-7-(2-phenylethyl)pyrrolo[2,3-d]pyrimidine (compound 1369) was the best inhibitor of cdk1 and cyclin B, with a 50% inhibitory concentration (IC₅₀; 14 μM) similar to that of roscovitine; it was competitive with respect to ATP (K_i = 14 μM). The potency of compound 1369 against cdk1 and cyclin B was similar to its cytotoxicity (IC₅₀s, 32 to 100 μM) but not its antiviral efficacy (IC₅₀s, 0.02 to 0.3 μM). Thus, our results indicated the null hypothesis. In contrast, roscovitine was only weakly active against HSV-1 (IC₅₀, 38 μM) and HCMV (IC₅₀, 40 μM). These values were similar to those derived by cytotoxicity and cell growth inhibition assays, thereby suggesting that roscovitine is not a selective antiviral. Therefore, we propose that inhibition of cdk1 and cyclin B is not responsible for selective antiviral activity and that pyrrolo[2,3-d]pyrimidines constitute novel pharmacophores which compete with ATP to inhibit cdk1 and cyclin B.