This review of the part played by the singlet states of molecular oxygen in the environment deals with atmospheric aspects. There are five bound excited states of molecular oxygen that correlate with two ground state, 3P, oxygen atoms. Of these, three are singlets, although the other two states (triplets) are closely associated with singlet oxygen processes, especially in the mesosphere. A weakly bound quintet state has been invoked, as well, in explaining some aspects of the physical chemistry of the singlet species. Of the three singlet states, the a1Δg is the most familiar. It has a low excitation energy, a long radiative lifetime, and is rather resistant to collisional deactivation in the gas phase. As a consequence, its chemistry has been susceptible to detailed study in the laboratory. These investigations, coupled with estimates of production rates, suggest that O2(a1Δg) is probably not important in initiating much chemical change in the lower atmosphere, at least in the gas phase; excited molecules dissolved in water droplets may promote chemical change under special circumstances. In the stratosphere and mesosphere, each of the bound excited states gives rise to characteristic emission features of the airglow, both by day and by night. The observational data, obtained from the ground, and from balloons, high-flying aircraft, rockets and satellites is surveyed as a background to examining the chemical and photochemical mechanisms by which the different states become excited. These mechanisms clearly differ by day and by night, and they also depend on the altitude from which the emission comes. The most intense feature of the oxygen dayglow, the Infrared Atmospheric Band, comes from O2(a1Δg) that is produced in the photolysis of ozone. Because dayglow measurements are sometimes used to derive ozone concentrations and altitude profiles in the atmosphere, the efficiency of production of the species in the photolysis of ozone is examined critically, and some unexpected laboratory findings are reported. The b1Σ+g state of oxygen is excited during the day largely by resonance scattering, although some is also populated by energy transfer from O(1D) to O2. At night, recombination of O(3P) atoms is the most likely source of excitation of all the states of oxygen. Laboratory experiments that bear on these processes are reviewed, and theoretical estimates of the partitioning of recombination events between the different states are presented. Direct recombination into the a1Δg and b1Σ+g states is unlikely to be efficient enough to produce the observed concentrations of these species, and some indirect process is thus implicated. Laser excitation experiments show that quenching of the three higher excited (ungerade) states of oxygen by O2 and, especially, N2, can generate O2(b1Σ+g) with high efficiency; similar experiments demonstrate explicitly that the quenching of O2(b1Σ+g) by the atmospheric gases yields O2(a1Δg). A consistent excitation scheme for the nightglow emissions is presented; this scheme also pays attention to the “auroral green” line produced by the 1S state of atomic oxygen, the intensities of which in the atmospheres of Earth and Venus provide some clues about the excitation of the molecular states. Finally, the laboratory studies are shown to indicate that the formation of excited molecular oxygen from vibrationally rich hydroxyl (OH) radicals is unlikely to be of major importance in the atmosphere.