In a clinical trial of age-related macular degeneration (AMD), the outcome measure chosen to assess the efficacy of treatment should reflect the purpose of the trial and the stage of development of the treatment. This article considers 3 classes of outcomes: continuous variables, such as mean change in best visual acuity; binary (2-category) variables, such as experiencing a 15-letter loss; or 3-category variables, such as experiencing either a 15-letter loss or 15-letter gain. Each type of outcome has advantages and disadvantages. Trials using outcomes based on means require much smaller sample sizes than trials based on 2- or 3-category variables, but means do not address the experience of individuals. Two- and 3-category variables show what happens to individuals, but they are subject to misclassification and are statistically inefficient. The authors recommend considering continuous measures for early stage trials and for trials studying various dose regimens when a treatment has been well characterized. However, 2- and 3-category outcomes are particularly useful in confirmatory phase 3 trials of a new therapy. A new graphical method is proposed to provide insight into the distribution of the time course of changes in acuity on an individual patient basis.