The production of mycobactin (MBT) by Mycobacterium tuberculosis is essential for this bacterium to access iron when it is in an infected host. Due to this essential function, there is considerable interest in deciphering the mechanism of MBT assembly, with the goal of targeting select biosynthetic steps for antituberculosis drug development. The proposed scheme for MBT biosynthesis involves assembly of the MBT backbone by a hybrid nonribosomal peptide synthetase (NRPS)/polyketide synthase (PKS) megasynthase followed by the tailoring of this backbone by N(6) acylation of the central l-Lys residue and subsequent N(6)-hydroxylation of the central N(6)-acyl-l-Lys and the terminal caprolactam. A complete testing of this hypothesis has been hindered by the inability to heterologously produce soluble megasynthase components. Here we show that soluble forms of the NRPS components MbtB, MbtE, and MbtF are obtained when these enzymes are coproduced with MbtH. Using these soluble enzymes we determined the amino acid specificity of each adenylation (A) domain. These results suggest that the proposed tailoring enzymes are actually involved in precursor biosynthesis since the A domains of MbtE and MbtF are specific for N(6)-acyl-N(6)-hydroxy-l-Lys and N(6)-hydroxy-l-Lys, respectively. Furthermore, the preference of the A domain of MbtB for l-Thr over l-Ser suggests that the megasynthase produces MBT derivatives with β-methyl oxazoline rings. Since the most prominent form of MBT produced by M. tuberculosis lacks this β-methyl group, a mechanism for demethylation remains to be discovered. These results suggest revisions to the MBT biosynthesis pathway while also identifying new targets for antituberculosis drug development.