Until very recently, it had not been possible to determine the structures of oxymyoglobin and oxyhaemoglobin, because the crystals autoxidized. My colleagues S. E. V. Phillips and B. Shaanan have been able to overcome this problem. Phillips did an X-ray analysis of sperm whale oxymyoglobin at 1.6 Å resolution, followed by a neutron diffraction analysis at 1.7 Å resolution in collaboration with B. Schoenborn. Shaanan solved the structure of human oxyhaemoglobin at 2.1 Å resolution. These analyses have provided firm stereochemical data about the conformation of the haem and the geometry of the FeO 2, bond. They have also revealed the role of the distal histidine in discriminating between O 2 and CO. Bent geometry for the FeO 2 bond with an FeOO angle of 114° was predicted by Pauling. This prediction was confirmed by X-ray studies of J.P. Collman's oxygenated picket fence complexes, except that the FeOO angle there was 130°. Intuitively, one would have expected that angle to be determined by the nature of the FeO 2 bond, and to remain the same in all oxygenated haem derivatives, but this is not true. The FeOO angle is 115(±5)° in oxymyoglobin and 156(±10)° in the α and β oxyhaemoglobin. The difference is due to the different constraints imposed by the distal residues in the haem pocket (His E7, Val E11 and Phe CD1). Collman's picket fence complexed with dimethylimidazole has a partition coefficient between O 2 and CO of 4280; the same complex with a covalently attached imidazole has a partition coefficient of 26,600. By contrast, the partition coefficients of myoglobin and haemoglobin are 150 and 250 respectively. If they were as high as in the picket fence complex, respiratory transport by haem proteins would not be possible, since CO is produced endogenously in the breakdown of porphyrin (one mol CO per mol porphyrin). How do the two proteins discriminate between O 2 and CO? The electronic structures of the two ligands ensure that O 2 binds preferentially in the bent conformation, while CO prefers to bind linearly with FeCO on the haem axis. The distal pockets in the myoglobin and haemoglobin are tailored so as to fit the bent oxygen, but to oppose the binding of the linear CO which is forced off the haem axis by steric hindrance. This appears to be one of the discriminating devices. The other consists of hydrogen bonding by the distal histidine. Solvent effects and spectroscopic evidence suggested that the FeO 2 bond is polar, with transfer of negative charge from the iron to the oxygen, while the FeCO bond is purely covalent. Pauling first suggested that the distal histidine could form a hydrogen bond to the terminal oxygen which carries a formal negative charge in his view of the FeO 2 complex. Evidence suggesting such a bond comes from electron paramagnetic resonance and oxygen affinity data on cobalt-substituted haemoglobins and myoglobins. Phillips and Schoenborn have now proved the existence of that hydrogen bond by a neutron diffraction analysis of oxymyoglobin. In D 2O at pH 8.4 the distal histidine carries one exchangeable deuteron which can bind either to N ϵ facing the bound haem ligand or to N δ, facing the external solvent. The neutron maps showed that in oxymyoglobin the deuteron is on N ϵ forming a hydrogen bond with the bound oxygen, while in carbonmonoxymyoglobin it is on N δ, facing the solvent, and the histidine is further removed from the ligand. This result confirms the polar character of the FeO 2 bond and the non-polar character of the FeCO bond. It also tells us that nature employs not one but two methods to discriminate between the two ligands. Eisenberger et al. used extended X-ray absorption fine structure (EXAFS) to measure the FeN distance in deoxyhaemoglobin and concluded that the irons lie only 0.2 +0.1 −0.2 Å from plane of the porphyrin nitrogens and that the cooperative mechanism I had proposed is therefore invalid. We have now compared the EXAFS of deoxyhaemoglobin with that of the ferrous ‘picket fence’ 2-methylimidazole complex in which the displacement of the iron from the plane of the porphyrin nitrogens is known to be 0.399 ± 0.004 and 0.426 ± 0.004 Å from the mean porphyrin plane. The two EXAFS spectra are very similar, consistent with similar displacements of the irons. We find the same FeN distance of 2.06 ± 0.01 Å in deoxyhaemoglobin as Eisenberger et al., but show that the displacement of the iron cannot be calculated from that distance.