RTrees::setCalculateVarImportance(true) has impact on training?

Hi all My understanding of calculating the variable importance is that this is a diagnostic feature only. However, I get reproducibly different training results if it is activated. Am I wrong? Thanks Kuno