A 9000-Mr protein isolated from a 60% ethanolic extract of soybean (Glycine max) seeds has been characterized and fully sequenced. The protein consists of 80 amino acid residues with four disulfide bonds. It contains a large number of hydrophobic residues and lacks methionine, phenylalanine, tryptophan, lysine and histidine residues. The protein readily crystallizes from water but is quite soluble in aqueous organic solvents like 95% 1-propanol. It aggregates to form large molecules (above 80 kDa) under ordinary denaturing conditions, such as 6 M guanidine · HCl and 8 M urea. Sequence analysis showed that the amino-terminal four-fifths is extremely hydrophobic and most of the acidic residues exist as their amide forms, and only the carboxyl-terminal short segment is rather hydrophilic. A computer search for homology detected an unexpected similarity of this protein to rat prolactin; however, its significance could not be assessed and this protein appears to represent a hitherto unknown protein family. Although no biochemical activity could be detected, the existence in relatively high abundance (approx. 200 mg from 1 kg seeds) of this novel protein may suggest its physiological significance in the plant.