Kovats Retention Index data for volatile compounds are normalized for the simple hydrocarbons such that pentane is 500, hexane 600, heptane 700, etc, and all retentions are relative to this. This way if carrier gas flow rates change, or temperature ramp changes, the change in retention times should allow recalculation quite easily using Kovats. Different columns will have different chemistries so this will be the biggest factor for the accuracy of retention time relative to retention index. Another factor is the size of the peak. Small peaks tend to be very close while larger peaks (that can have tailing) can be off +/- up to 20 index points. DB-1 (100% PDMS) and DB-5 (95% PDMS/5% Phenyl) seem to work best for Kovats since generally very hydrophobic columns have simple, preictable chemistry (less interaction/more boiling point dependent).
This is a good website with many compounds Kovats' indices:
http://www.flavornet.org/flavornet.htmlAlso, this article link below shows that for four types of GC columns (OV-101, DB-1, DB-5, DB-Wax) you can predict the Kovats' index for a volatile compound using only the partition coefficient (Kow) and the boiling point to an R^2>0.997 for OV-101, DB-1 and DB-5 and R^2>0.982 for DB-Wax. Quite useful even if you have a mass spec and get a poor match that doesn't make sense, this can usually help eliminate that compound. Generally boiling point and partition coefficient can be found in literature, but for rarer compounds BP and Log P can be predicted using chemspider typing in the SMILES or InChi notations.
http://naldc.nal.usda.gov/download/17239/PDF