This study examined the test-retest reliability of the error-related negativity (ERN) and error positivity (Pe) amplitudes using a Flanker task in 118 neurotypical children and 53 adults before and after latency jitter adjustments. The reliability of the ERN and Pe amplitudes was moderate for children and moderate to strong for adults. The latency variability adjustment did not improve the reliability of the ERN and Pe amplitudes for either group, suggesting that latency variability may be a trait-like measure. For comparison purposes, the reliability of the stimulus-locked ERPs was strong for correct trials, yet the reliability was weak for incorrect trials.