Major discoveries have been made in the past few years on the structure and mode of biosynthesis of serum albumin. The complete amino acid sequence of this protein has been determined, and its covalent structure shown to be a single peptide chain grouped into a series of nine disulfide-bonded loops. These loops appear to associate into three similar domains. By study of isolated fragments of the molecule it can be demonstrated that the binding of billirubin and the primary binding of long-chain fatty acids are functions of separate domains. The biosynthesis of albumin has been found to involve a precursor form, termed "proalbumin", in which a basic hexapeptide is attached to the amino end of the chain. Similar precursor forms are now known to have a role in the formation of other secreted proteins, but in the case of albumin the purpose of the additional peptide is not clear. Clinical methodology for albumin assay has advanced but little despite--or perhaps in part because of--the increasing use of automation. Hope for improvement is foreseen in the advent of immunochemical procedures and in a better understanding of the specificity of dye-binding reactions.