"Theorem A" in this paper by Y. Gordon: 
http://www.math.uiuc.edu/~mjunge/Comps/Gordonm.pdf
is a comparison inequality for Gaussian processes. 

Is there an analogue of this result for subgaussian processes?