Most of the immunoreactive growth hormone (IRGH) in human plasma elutes from Sephadex G-75 as "little" GH (LGH), mol wt 22,000, but 14-39% elutes earlier ("big" GH, BGH). In saline extracts of human pituitary, 11-17% of IRGH eluted as BGH. On gel filtration of pituitary and plasma BGH in 8 M urea, 59-81% ran as LGH, but when the remaining BGH was refiltered in urea, all ran as BGH. Thus there is a "urea-stable" and a "urea-labile" form of BGH. SImilarly, freezing and thawing converted over half of pituitary and plasma BGH to LGH, but when the "freeze-stable" BGH was again frozen, thawed, and refiltered, almost all ran as BGH. Urea-stable BGH was not dissociated by freezing, and most of the freeze-stable BGH was stable in urea, so the two forms are very similar or identical. Since 8 M urea and freezing dissociate peptides linked by noncovalent bonds, probably the BGH that is dissociated by urea and freezing consists of LGH bound noncovalently to another moiety, while in stable BGH the LGH is bound to another molecule by covalent or unusually strong noncovalent linkage. On centrifugation, the sedimentation of urea-stable BGH was consistent with a mol wt about twice that of LGH. Trypsinization of urea-stable BGH converted 36-59% to LGH, suggesting that some BGH may be a "prohormone" of LGH. On retrypsinization of the BGH that was not converted to LGH, only 13-24% converted, suggesting that there may be two forms of urea-stable BGH which vary in their response to trypsin.