A retrospective multivariate analysis of 264 consecutive cases of vitrectomy for removal of epiretinal membranes of the macula was performed. Two main preoperative prognostic factors were identified that were associated with the visual outcome of surgery: (1) the preoperative visual acuity, and (2) the duration of blurred vision before surgery. Discriminant functions constructed using these two factors allowed correct classification of eyes into an outcome group with good visual acuity with a 69% accuracy, and an 80% accuracy in classifying eyes into a group showing significant improvement in the level of visual acuity. Eyes that began with vision better than 20/100 tended to have slightly better postoperative visions than eyes with poorer preoperative visual acuity. However, eyes with worse preoperative vision tended to improve the greatest amount after surgery. Eyes with a longer history of blurred vision had poorer visual acuity after surgery and also less improvement in vision. Other statistically significant prognostic factors were identified; however, they did not appreciably improve the accuracy of the discriminant function in predicting visual outcome.