Published and newly calculated pA2-values of 147 neurohypophyseal hormone analogues (7 positions varied) acting as inhibitors of oxytocin on isolated rat uterus in vitro have been subjected to fractionation according to the method by Free and Wilson which was slightly modified for this purpose. The computation was carried out in several steps. After each step, substances with outlying pA2-values were eliminated. The reduced group containing 73-79% of the original substances displayed a high degree of additivity of side chain contributions (SCC). This group seems to follow the "participation" rule as formulated by Free and Wilson. Analysis of the group of eliminated substances and of the resulting SCC-spectrum (level diagram) enabled us to draw some conclusions concerning the structural requirements of receptor binding: i) The intact ring structure is necessary for the peptide-receptor interaction: linear peptides or peptides with an extended ring are always outliers; ii) Carba analogues (substitution with CH2 in the disulfide ring) display better affinities than peptides with an S-S ring; D-Arg8 substitution decreases the binding affinity; iii) Considerably better additivity is achieved when peptides are divided into subgroups with vasopressin-like and oxytocin-like features; populations of receptors more specific for vasopressin and for oxytocin, respectively, can be assumed. Estimates of the "true" receptor-peptide dissociation constants can be obtained by summation of the corresponding SCC's in each investigated position. The value obtained for oxytocin is identical with the medium affinity binding site on myometrial cells, and not with the high affinity site. A nonlinear relationship exists between SCC's computed from pA2-values for magnesium-free and magnesium-containing (0.5 mM) media but no evidence speaks in favor of a Mg-potentiating effect on receptor binding.