We present molecular mechanics simulations on covalent complexes between d[(GC)5]2, d(G10).d(C10), d(GCGCGAGCGC).d(GCGCTCGCGC), d(GCGCGTGCGC).d(GCGCACGCGC), d(G5AG4).d(C4TC5), and d(G5TG4).d(C4AC5) on one hand and potent antitumor antibiotics anthramycin and neothramycin A on the other, using the all atom force field in the framework of the program AMBER(UCSF). The energy-refined models of both the sets of complexes show minimal distortions for the nucleotides, consistent with the results of 2D NMR studies on these complexes. The drugs have 3'-orientation in the minor groove, consistent with the previously reported investigations employing the united atom force field and with the experimental observations. Both anthramycin and neothramycin are calculated to bind preferentially to the puGpu sequences over pyGpy. This is in qualitative agreement with experimental studies for anthramycin, while for neothramycin A, this result is in apparent disagreement with experimental observations which have reported preferential binding of neothramycin A to poly(dG-dC).poly(dG-dC) over poly(dG).poly(dC). While the present study brings out the usefulness of the simple molecular mechanics approach (using an all atom force field) in rationalizing substantial experimental observations, it also emphasizes the need for further investigations on solvent and dynamics effects in understanding the sequence specificity of drug-DNA binding.