Purpose: To compare different reading speed measures for capturing longitudinal visual function changes in eyes with geographic atrophy (GA) secondary to age-related macular degeneration. Design: Analysis of data from Chroma (NCT02247479) and Spectri (NCT02247531), 2 identically designed, phase III, double-masked, randomized controlled clinical trials for lampalizumab. Participants: Nine hundred forty participants aged >= 50 years old with bilateral GA, who completed monocular testing of reading speed at >= 3 visits over >1-year follow-up. Methods: Monocular reading speed was assessed using the Minnesota Low-Vision Reading Test (MNRead). Four different reading speed measures were derived and compared: reading speed of the fastest sentence read (RS1), mean reading speed of the 3 fastest sentences read (RS2), mean reading speed of the sentences larger than the critical print size (RS3), and mean reading speed of the 10 largest print sizes (termed the Reading Accessibility Index [ACC]). Main Outcome Measures: Coefficient of variation (CV), with lower values reflecting better performance of a measure for capturing longitudinal change relative to interindividual variability. Results: All 4 reading speed measures showed a significant decline at 48, 72, and 96 weeks from baseline (P < 0.001 for all). The CVs for ACC and RS2 (204% and 208%, respectively) were lower than for RS1 (255%; P <= 0.002) and RS3 (224%; P >= 0.068) for detecting change from baseline at 48 weeks, but these 2 measures were not significantly different from each other (P = 0.362). There were also statistically significant, but weak, negative correlations between the change from baseline at 48 weeks for all 4 reading speed measures with GA area on fundus autofluorescence imaging (rho = -0.13 to -0.15; all P < 0.001). Conclusions: The mean reading speed derived from either the 10 print sizes found in everyday life (ACC) or the fastest 3 sentences read (RS2) was better than 2 widely used measures (RS1 and RS3) at capturing progressive functional decline in eyes with GA and may be the preferred measures in future clinical trials and studies. All reading speed measures also showed an expected progressive decline over time, but they only showed a weak correlation with GA growth.