The proteins ESAT-6 and CFP-10 have been shown to be secreted by Mycobacterium tuberculosis and Mycobacterium bovis cells, to be potent T-cell antigens, and to have a clear but as yet undefined role in tuberculosis pathogenesis. We have successfully overexpressed both ESAT-6 and CFP-10 in Escherichia coli and developed efficient purification schemes. Under in vivo-like conditions, a combination of fluorescence, circular dichroism, and nuclear magnetic resonance spectroscopy have shown that ESAT-6 contains up to 75% helical secondary structure, but little if any stable tertiary structure, and exists in a molten globule-like state. In contrast, CFP-10 was found to form an unstructured, random coil polypeptide. An exciting discovery was that ESAT-6 and CFP-10 form a tight, 1:1 complex, in which both proteins adopt a fully folded structure, with about two-thirds of the backbone in a regular helical conformation. This clearly suggests that ESAT-6 and CFP-10 are active as the complex and raises the interesting question of whether other ESAT-6/CFP-10 family proteins (22 paired genes in M. tuberculosis) also form tight, 1:1 complexes, and if so, is this limited to their genome partner, or is there scope for wider interactions within the protein family, which could provide greater functional flexibility?