A number of analogues of ovine growth hormone (GH), in which regions of the hormone had been deleted, were produced by site-directed mutagenesis, and characterised by radioimmunoassays and radioreceptor assays. These analogues were based on a previously described variant (oGH1) in which an 8-residue extension replaces the N-terminal alanine of pituitary-derived ovine GH. Three analogues with deletions near the N-terminus were studied, with shorter extensions of 7 or 1-2 residues (oGH14, oGH5) or with the N-terminal sequence Ala-Phe-Pro- of pituitary-derived ovine GH replaced by Thr-Met-Ile-Thr- (oGH11). These modifications had little effect on potency in radioimmunoassays based on a polyclonal antibody and five different monoclonal antibodies (MABs), or in a radioreceptor assay, indicating that the N-terminal sequence was not included in the epitope binding to any of the monoclonal antibodies, or a major epitope binding to the polyclonal antibody, or in receptor binding site 1. A variant in which residues 133-139 were deleted retained full binding to 4 of the 5 MABs, suggesting correct folding, but markedly reduced binding to MAB OA16, suggesting that the epitope for this MAB includes some or all of these residues. This variant also failed to displace about 35% of labelled hormone from the polyclonal antibody studied, suggesting that residues 133-139 may be involved in a major epitope for this antibody. This variant showed slightly lower receptor binding activity than ovine GH. Two other deletion variants - oGH1Delta33-46 (equivalent to the naturally occurring 20K variant of human GH) and oGH1Delta180-191 (lacking the C-terminal 12 residues) showed poor folding efficiency and solubility, and low binding to all MABs except OA15, which has a linear epitope. The results suggest that these variants were incorrectly folded, but interestingly they did retain some activity in the receptor-binding assay (respectively about 5% and 0.5% of the activity of ovine GH itself).