1. The receptors for relaxin in the rat atria and uterus were investigated and compared by use of a series of synthetic and native relaxin analogues. The assays used were the positive chronotropic and inotropic effects in rat spontaneously beating, isolated right atrium and electrically driven left atrium and the relaxation of K+ precontracted uterine smooth muscle. 2. Relaxin analogues with an intact A- and B-chain were active in producing powerful chronotropic and inotropic effects in the rat isolated atria at nanomolar concentrations. Single-chain analogues and structural homologues of relaxin such as human insulin and sheep insulin-like growth factor I had no agonist action and did not antagonize the effect of the B29 form of human gene 2 relaxin. 3. Shortening the B-chain carboxyl terminal of human gene 1 (B2-29) relaxin to B2-26 reduced the activity of the peptide and removal of another 2 amino acid residues (B2-24) abolished the activity. This suggests that the B-chain length may be important for determination of the activity of relaxin. More detailed studies are needed to determine the effect of progressive amino acid removal on the structure and the bioactivity of relaxin. 4. Porcine prorelaxin was as active as porcine relaxin on a molar basis, suggesting that the presence of the intact C-peptide did not affect the binding of the prorelaxin to the receptor to produce functional responses. 5. Relaxin caused relaxation of uterine longitudinal and circular smooth muscle precontracted with 40 mM K+. The pEC50 values for human gene 2 and porcine relaxins were lower than those in the atrial assay, but rat relaxin had similar pEC50 values in both atrial and uterine assays. Rat relaxin was significantly less potent than either human gene 2 or porcine relaxin in the atrial assay, but in the uterine assay they were equipotent. The results suggest that the relaxin receptor or the signalling pathway in rat atria may differ from that in the uterus.