Objective: Although the Immediate Post-Concussion Assessment and Cognitive Test (ImPACT) is the most widely used computerized neuropsychological battery for sport-related concussion, reliability findings vary and research regarding its factor structure is limited. The purpose of this study was to examine ImPACT’s retest reliability using a two-factor model proposed by Schatz and Maerlender (2013). Method: Participants included 109 NCAA Division III male and female athletes who did not have a concussion and volunteered to undergo ImPACT retesting from 24 hours to 4.6 years after baseline. A Memory factor score was the average of Verbal and Visual Memory composite scores and a Speed factor score was the average Visual Motor Speed and Reaction Time composites. Results: Based on Slick’s (2006) guidelines, intraclass correlation coefficients (ICCs) ranged from low to high for the following retest intervals from baseline: to 12 months (n = 54, ICC = .46–.81), to 12–24 months (n = 20, ICC = .77–.81), and to over 24 months (n = 35, ICC = .7–.72). Conclusions: These results are comparable to the two-factor ICC values reported in Schatz and Maerlender but higher than the ICC values for ImPACT’s four composite scores from the same sample (ICCs = .3–.81).