Locally grown kumara (sweet potato; Ipomea batatis) was used as the starting point for the purification of a proteinase inhibitor. The purified inhibitor was highly specific for trypsin, and much less effective as an inhibitor of chymotrypsin. Two 22kDa variants were present, closely homologous to each other and to sporamin A, but with a single amino acid substitution (proline in place of serine, the second residue in mature sporamin A). One variant had the same N-terminus as sporamin A, whereas the other had a tripeptide N-terminal extension, which may represent an intermediate in the proteolytic processing of the precursor protein. A larger variant was apparently a disulphide-linked dimer of the monomeric inhibitor. A rabbit polyclonal antiserum prepared against the trypsin inhibitor reacted with all of these variants, but did not cross-react with commercially-available soybean proteinase inhibitors. The purified inhibitor did resemble other proteinase inhibitors in having a biphasic effect upon the proliferation of human fibroblasts, with a mitogenic action at low concentrations, and an inhibitory effect at higher concentrations.