Data from a selected sample of 158 patients who received excimer laser photorefractive keratectomy (PRK) in our center and obtained a desirable refractive outcome at 6-months were analyzed retrospectively to establish an empirical model of refractive power after PRK. This model, generated by a computer-assisted regression program, is presented as a multivariate non-linear quadratic equation. Since the coefficients of each variable (attempted correction and postoperative days) are statistically significant, the estimated influence of the variables on the postoperative refraction from this regression model is reliable and consistent. Therefore, this model has been used to direct the use of corticosteroids for modulating post-PRK refraction. The predictability of this model has also been corroborated by separate data from 1-year follow up of PRK recipients. The recipients whose refraction at 6 months agree with the calculation from this model have obtained an excellent outcome; while those whose refractive outcome deviates from this model have gotten undesirable results.