The present research is focused on the introduction of a new optimization parameter in the field of comprehensive 2D GC (GC×GC), herein defined as "gas velocity at the point of re-injection". GC×GC experiments were performed using a loop-type cryogenic modulator and a rapid-scanning quadrupole mass spectrometer, as detection system. All experiments were performed using a 30m×0.25mm ID apolar primary column, and a 1m×0.10mm ID medium-polarity secondary one. With regards to the modulator loops, three types of uncoated columns were used, namely with a 0.25, 0.18, and 0.10mm ID. It was found that under the same second-dimension analytical conditions (modulation, oven temperature, gas velocity) the quality of the GC×GC separation was dependent on the gas velocity at the moment of re-injection. For a loop-type modulator, the re-injection point is located at the downstream cooling/heating point of the delay loop. It was found that the GC×GC performance gradually improved when the ID of the modulator loop decreased (and the re-injection gas velocity increased), and reached its best when the 0.10mm ID uncoated column was used. GC×GC experiments were performed on a C14 n-alkane (to evaluate the efficiency of band re-injection, in terms of peak widths) and on an essential oil sample (to evaluate peak widths, resolution and the overall separation performance). The results herein reported are valid for practically any type of cryogenic modulator.