It depends on the sequence of the viral transcript. If there are some poly-(A) stretches an oligo(dT) primer will work as good as random hexamers.
When I investigated the expression of a comovirus in soybean I used oligo(dT) primers as well as gene specific primers and random hexamers for the RT-reaction. All combinations worked properly. Since the RT reaction is performed at low temperatures, small poly (A) stretches might be sufficient for the stable binding of the oligo (dT) primer.
Reply to Manuel Müller: There is no problem when just clone some viral transcripts. However, when we do qRT-PCR, each viral transcript has its own diverse small poly (A) stretches, so the efficiency of RT-PCR is dependent on its stretches. That's the problem!
Okay, now i understand your issue. If you do proper normalization of your qPCR data it doesn´t matter whether the efficiency of the RT reactions is different for the transcripts.
The other alternate would be to standardize a primer that works for all the the transcripts, you may have to find a fairly conserved region in the genome. However above suggestion of proper normalization should work equally good.