Many different compounds bind to human serum albumin (HSA), which can be a significant problem in the drug discovery process. To aid in the design of drug molecules that do not bind to HSA, the structures of HSA/ligand complexes would be very useful. However, little information has been reported on the structures of small molecules complexed to HSA. In this paper, we describe a procedure for preparing isotopically labeled domains of HSA for nuclear magnetic resonance (NMR) studies. The procedure involves the expression in Escherichia coli refolding, and a multistep purification. Domains I and III are capable of folding into stable structural units and producing well resolved N-15/H-1 correlation spectra, whereas domain II forms significant aggregates at sub-millimolar concentration. Using our protocols, isotopically labeled and properly folded domains I and III can be effectively produced in large quantities for NMR-based structural studies and NMR-based screening. This provides a valuable tool for obtaining structural information on HSA/ligand complexes by NMR which will be useful in drug discovery.