Objective: To compare the validity indices of the King-Devick (KD) test in hockey players using any increase in test time over baseline to a 6-second increase as a positive concussion test. We hypothesized the KD test using the 6-second change would yield greater validity indices. Design: Prospective observational cohort study. Setting: Sports complex. Participants: Sixty-five male hockey players aged 13 to 20 years. Independent Variable: Concussion diagnosis. Main Outcome Measures: Time to complete the KD test. Sensitivity (SN), specificity (SP), and the positive likelihood ratios (LRs+) were calculated using 2 thresholds for a positive test: (1) any increase in time and (2) an increase of 6 seconds or greater. Results: Eighteen players (27.7%) were diagnosed with a concussion. Using any increase in time as a positive test yielded the following: SN 72.2%, SP 78.7%, LR+ 3.4, and LR− 0.4. Using the 6-second threshold yielded the following: SN 44.4%, SP 93.6%, LR+ 7.0, and LR− 0.6. A receiver operator curve analysis confirmed 6.40 seconds or greater maximized the LR+. Conclusions: A 6-second or greater increase in the KD test performance provides greater validity in diagnosing concussion as compared with any increase in performance time.