Few data exist from which an optimal strategy for retinopathy of prematurity (ROP) screening can be deduced. This strategy should compromise between the dual purpose of screening, to wit monitoring the incidence of ROP and determining an optimal moment for therapeutic intervention. We recorded the timely incidence and course of ROP in all except four premature infants admitted during a 1-year period. The study results indicate that a single screening moment is not likely to detect more than 70% of all cases. When relying, however, on a single screening, the investigation should be performed between the 7th and 9th weeks. The unpredictable course of ROP, varying between early starting, rapidly progressing forms and very late starting, usually benign forms probably precludes the possibility of 100% ROP detection. Screening during the 4th to 5th, 8th, and 11th weeks probably gives the best chances for determination of an optimal moment for treatment and approximation of the incidence.