This study examined within-session reliability and minimal detectable change (MDC) in mean split time, velocity, and acceleration across 100-meter start phases (0-20, 20-50, 50-70, 70-100 m) in elite speed skaters. Nineteen skaters (10 females, 9 males) completed two trials on the same day on a 400-meter indoor ice track under standardized conditions, with split times recorded every 10-meters using an automated timing system. Results showed acceptable reliability for mean split time (ICC = 0.84, CV = 7.5%, MDC = 0.00-0.15 s) and velocity (ICC = 0.79, CV = 9.2%, MDC = 0.21-1.09 [metres per second]). Mean acceleration showed moderate reliability (ICC = 0.70, CV = 12.4%) and higher variability over greater distances (70-100 m). These findings support using the 100-meter acceleration profile to assess key performance elements (initial start and sustained acceleration), emphasizing the need to tailor training and assessments to specific race phases.