The affinity of 22,000-Mr human growth hormone (22 K-hGH) for GH binding proteins in rabbit liver is increased approx. 19-fold by 25 mM-Ca2+. In contrast, ovine growth hormone (oGH) binding is Ca2+-independent up to 10 mM, and decreased by greater Ca2+ concentrations. The 20,000-Mr hGH variant (20K-hGH), lacking residues 32-46, exhibits intermediate behaviour. Without Ca2+ there is a residual 40% of maximum specific binding to liver microsomes, and this increases to 65% with liver cytosolic GH binding proteins. In contrast with 22K-hGH, Scatchard analysis of 20K-hGH binding to liver microsomes produces curvilinear plots in the presence of 25 mM-Ca2+. From these results and inhibition studies with monoclonal antibodies to the GH binding proteins, it is concluded that deletion of the region 32-46 from 22K-hGH has eliminated one component of high-affinity Ca2+-potentiable binding. The Ca2+-mediated increase in Ka for the 22K-hGH-binding protein interaction is consistent with convergence of unit negative charges on the hormone and binding protein towards an intercalated Ca2+ ion. A positive charge in the critical region of nonprimate GHs would render their interactions Ca2+-independent and of lower Ka compared with 22K-hGH. A likely candidate for the negatively charged interactive residue is glutamate-33, since it is unique to human GH and is replaced by a positively charged arginine in non-primate GHs. Its absence in 20K-hGH could explain the altered calcium-dependence of 20K-hGH binding to what is probably the type 2 binding protein [Barnard & Waters (1986) Biochem. J. 237, 885-892]. The Ca2+-dependence of 20K-hGH binding to a subset of GH binding proteins provides both a verification and a mechanistic basis for the proposal [Hughes, Tokuhiro, Simpson & Friesen (1983) Endocrinology (Baltimore) 113, 1904-1906] that 20K-hGH binds with high affinity to only a subset of binding proteins in rabbit liver membranes.