Purpose: To compare the outcome of early treatment at the initial stage 3 moderate phase of retinopathy of prematurity (ROP) with that at prethreshold ROP. Methods: The outcome of treatment of 117 very-low-birth-weight infants was reviewed. Sixty-eight infants underwent their first treatment at prethreshold ROP (group A). The other 49 infants underwent their first treatment at the initial stage 3 moderate phase of ROP (pre-prethreshold ROP) (group B). The first treatment was performed by laser photocoagulation or cryotherapy. Eleven clinical variables were included in the evaluation of the two groups. The outcome of treatment of the two groups was compared as favorable or unfavorable. Logistic regression analysis was performed to assess the relation between each independent variable and an unfavorable outcome. Results: The extent of contiguous stage 3 moderate ROP and that of cumulative stage 3 moderate ROP were significantly smaller in group B. There were nine unfavorable outcomes in group A, but none in group B. Logistic regression analysis showed that the extent of cumulative stage 3 moderate ROP and the presence of zone 1 retinopathy were major predictive factors for unfavorable outcome in the treatment of very-low-birth-weight infants. Conclusion: Early treatment at the initial stage 3 moderate phase of ROP decreases the occurrence of severe retinal changes. Purpose: To compare the outcome of early treatment at the initial stage 3 moderate phase of retinopathy of prematurity (ROP) with that at prethreshold ROP. Methods: The outcome of treatment of 117 very-low-birth-weight infants was reviewed. Sixty- eight infants underwent their first treatment at prethreshold ROP (group A). ​​The other 49 infants underwent their first treatment at the initial stage 3 moderate phase of ROP (pre-prethreshold ROP) (group B). The first treatment was performed by laser photocoagulation or cryotherapy. Eleven clinical variables were included in the evaluation of the two groups. The outcome of treatment of the two groups was compared as favorable or unfavorable. Logistic regression analysis was performed to assess the relation between each independent variable and an unfavorable outcome. Results : The extent of contiguous stage 3 moderate ROP and that of cumulative stage 3 moderate ROP were significantly smaller in group B. There were nine unfavorable outcomes in group A, but none in group B. Logistic regression analysis showed that the extent of cumulative stage 3 moderate ROP and the presence of zone 1 retinopathy were major predictive factors for unfavorable outcome in the treatment of very-low-birth-weight infants. Early treatment at the initial stage 3 moderate phase of ROP decreases the occurrence of severe retinal changes.