The unique of CO-cytochrome oxidase as first noted by Yonetani et al. (22) is that after its photodissociation at low temperatures recombination occurs as the sample temperature is raised, but at temperatures considerably higher than those for other CO-heme and CO-hemoprotein complexes; that is, the half recombination temperature is 180 K contrary to 25-30 K for other CO complexes. The photodissociability, however, disappeared when monomeric cytochrome oxidase was treated with pCMB to remove an intrinsic copper, the significance of which in CO complex formation was thus demonstrated. It is proposed that the copper is situated close to heme a and traps the photodissociated CO. The access of the trapped CO to the heme a to resume the original binding is effected only when sufficient energy for thermal agitation is provided by elevating the sample temperature. During the course of this study, new photo- and thermochromic properties were observed with the reduced enzyme by cooling it in liquid nitrogen after preincubation at pH 8.6 to 10.5. The characteristic bands appeared at 575 and 428 nm and diminished when this ample was illuminated at 26 K. As the sample temperature was raised these bands were restored with a half transition temperature of 80 K. When the reduced oxidase had been complexed with CO, cyanide or azide, or treated with pCMB, such a unique species did not appear. The enthalpy change of 1.16 kcal/mol for the formation of this species as well as the above-described properties suggests that the hydrogen bond between the formyl side group of heme a and one of seven sulfhydryl groups in cytochrome oxidase is responsible for the appearance and disappearance of this new species. Based on these results a schematic model has been proposed for the photo- and thermochromism of cytochrome oxidase at cryogenic temperatures and for the microenvironment of the prosthetic heme a and copper in this enzyme. On the other hand, contrary to the central dogma of Warburg that all CO-heme and CO-hemoprotein complexes are photodissociable, we observed little photodissociability with some CO-heme complexes, especially at very low temperatures, and presented a view that depending on the bond type between CO and heme iron the efficiency of photodissociation is so varied that under certain conditions practically no photodissociation occurs. According to this view a tilted arrangement of the ligated CO towards the heme plane accompanying a large extent of overlapping of the dpi(Fe) and the pi* antibonding orbital on the CO facilitates photodissociation. In addition to our own observations of photochemical properties of cytochrome oxidase and heme model compounds, recent photodynamic studies carried out by other investigator on CO-heme and CO-hemoproteins are summarized and the validity and limitation of their models are discussed.