A methodology has been developed for characterising hydrogen-bonded ring motifs formed between two organic molecules without any prior knowledge of the topology or chemical constitution of the motifs. The method has been implemented by modifying the current Cambridge Structural Database (CSD) System programs. All intermolecular ring motifs comprising ⩽20 atoms formed with N—H···N, N—H···O, O—H···N and O—H···O hydrogen bonds in organic structures in the CSD have been classified. The 75 bimolecular motifs occurring in >12 structures in the CSD are described in terms of their graph sets and chemical functionalities. Motifs are ranked according to their frequency of occurrence and according to their probabilities of formation, i.e. their frequency relative to the number of possible motifs which could have formed. These probabilities provide insights into the relative robustness of known and potential supramolecular synthons.