Hello FreeSurfers

I'm running some permutation tests and this induce me to a question,

I'm using mri_glmfit-sim as recommended with the mc-z method, running 1000 permutations and seeding randomly, I have ran this with the same data two different times, and the results are bit different, could affect the number chossen for the seed flag affecting the results, what this seeding flag does in the command?

Can any one point me to this?

Bests,
Gabriel