Retinopathy of prematurity (ROP) emerged as a cause of blindness in children in industrialized countries during the late 1940s and 1950s with increased survival of preterm babies due to improvements in neonatal care including use of supplemental oxygen. During the 1950s, ROP was the single commonest cause of blindness in children in many industrialized countries (the "first epidemic"). Hyperoxia was proposed as an important risk factor which was largely supported by laboratory-research and clinical studies. The use of oxygen was restricted in the mid-1950s which was followed by a reduction in the incidence of blindness from ROP, but higher rates of infant mortality and cerebral palsy. Oxygen was used more liberally in the 1960s, and blindness from ROP began to re-emerge (the “second epidemic”). The introduction of increasingly sophisticated technology, including accurate methods of monitoring oxygen, and better management of neonatal and perinatal complications in prematurely born infants in the 1970s were probably the major factors responsible for reduction of blinding ROP observed during this period. In industrialized countries, blindness from ROP is now largely restricted to infants in the extremely low birth weight group. However, ROP has been described with increasing frequency in regions with rapidly developing neonatal care and this has been termed the “third epidemic.” The reasons for this epidemic are mixed: premature birth and low birth weight dominate in settings which are able to deliver high-quality neonatal care while exposure to other potentially modifiable risk factors, including oxygen, being important in neonatal units which have shortages of resources and so are unable to provide optimum standards of care.