A serious problem in the replication of repeating-sequence DNa polymers using Escherichia coli DNA polymerase I arises from the fact that this polymerase has a very strong preference for the replication of poly[d(A-T)]. Thus reactions primed with DNA containing small amounts of contaminating poly[d(A-T)] will eventually result in complete domination of the synthesis by poly[d(A-T)]. This problem can be overcome by the addition to the reaction mixture of the synthetic quinoxaline antibiotic TANDEM which binds specifically to poly[d(A-T)] completely inhibiting its replication. Using thermal denaturation experiments it can be shown that TANDEM does not bind to most other synthetic DNA polymers (e.g., poly(dA) . poly(dT) and poly[d(A-T-C)] . poly[d(G-A-T)] and therefore their replication is not inhibited. The only exception we have encountered is poly[d(T-A-C)] . poly[d(G-T-A)] which does bind TANDEM and therefore the drug cannot be used during the synthesis of this polymer. The fact that poly[d(T-A-C)] . poly[d(G-T-A)] does bind TANDEM while poly[d(A-T-C)] . poly[d(G-A-T)] does not, suggests that the drug recognizes T-A rather than A-T sequences.