Paired-pulse transcranial magnetic stimulation (TMS) using fixed test stimuli suffers from marked variability of the motor evoked potential (MEP) amplitude. Threshold tracking TMS (TT-TMS) was introduced to overcome this problem. The aim of this work was to describe the absolute and relative reliability of short-interval intracortical inhibition (SICI) using TT-TMS. Cortical excitability studies were performed on twenty-six healthy subjects over three sessions (two recordings on the same day and one seven days apart), with MEPs recorded over abductor pollicis brevis. Reliability was established by calculating the standard error of the measurements (SEm), minimal detectable change (MDC) and intraclass correlation coefficient (ICC). Resting motor threshold and averaged SICI presented the lowest SEm and highest ICCs. SICI at 1”¯ms showed a higher SEm than SICI at 3”¯ms, suggesting different physiological processes, but averaging SICI over a number of intervals greatly increases the reproducibility. The variability was lower for tests undertaken at the same time of day seven days apart compared to tests performed on the same day, and in both instances the ICC for averaged SICI was â‰¥0.81. The MDC in averaged SICI was reduced from 6.7% to 2% if the number of subjects was increased from one to eleven. In conclusion, averaged SICI is the most reproducible variable across paired-pulse TT-TMS measures, showing an excellent ICC. It is recommended that, in longitudinal studies, testing be performed at the same time of day and that changes in cortical excitability should be measured and averaged over a number of interstimulus intervals to minimise variability.