Most organisms require dioxygen (O
2
) to survive; however, simple O
2
diffusion in blood does not supply a sufficient amount of oxygen for respiration. Thus, in all mammals, O
2
is carried and stored by O
2
‐binding heme proteins, myoglobin, and hemoglobin. Myoglobin is a single‐chain globular protein with heme as a prosthetic group, while hemoglobin consists of four heme‐containing subunits. O
2
binds to ferrous pentacoordinated heme‐iron and is stabilized by distal and proximal amino acid residues in the heme pocket. The O
2
pathway in the protein matrix and the binding mechanisms are discussed in detail by O
2
‐binding kinetic data for a series of myoglobin and hemoglobin mutants. Particularly, several residues, B10, E7, E10, E11, F7, and F8, mainly regulate the O
2
binding in the vertebrate heme proteins. In contrast, invertebrate and plant hemoglobins with unusually high O
2
affinities have been found in nature. In
Ascaris
hemoglobin, O
2
dissociation is very slow, whereas leghemoglobin shows a large O
2
association rate constant. In the case of trematode hemoglobin from
Paramphistonum epiclitum
, both kinetic parameters are improved. Artificial molecular design to modify O
2
affinity is another way to understand the O
2
binding mechanisms of the heme proteins. The strategy of heme protein modification can be divided into at least two approaches: One is a point mutation at distal or proximal amino acid residues, and the other is reconstitution by replacement of the native heme with an artificial heme. The physiological function of vertebrate hemoglobins is O
2
transport from the lungs or gills to the tissues; thus, they are required to exhibit cooperative interactions in binding O
2
to heme‐iron in the protein matrix. Although it is well known that a Hill coefficient of approximately 3.0 has been observed for tetrameric human hemoglobin, the mechanism of allosteric interaction is still discussed based on theoretical and experimental data. At present, an a two‐state (MWC) model proposed by Monod, Wyman, and Changeux for cooperative ligand binding is preferred to a sequential model. In addition, a study on the allosteric effectors for hemoglobins seems to be attractive from physiological aspects. The conventional effectors such as organic polyphosphate for human hemoglobin reduce the O
2
affinity of the T‐state, while it has been found that bezafibrate (BZF) or inositol hexaphosphate (IHP) can regulate O
2
binding not only for the T‐state but also for the R‐state. Finally, the interaction of NO with the heme proteins and modification of myoglobin for conversion into a new oxidase are briefly described.