I think this is a very valid question. From a backtesting point of view (& I assume this is where you are coming from, sorry if not....), overnight movements in US Index futures can sometimes be larger than expected ie quite important if stops are close. Any backtesting with OHLC bar data will require assumptions about movement within the bar, and particularly so if overnight data is excluded (nearly always the case it seems).
I have extracted 18 months back data from act forex's charts via www.gcitrading.com, also available as tables of data. The data includes Sunday evenings and all after hours. Supposedly this is obtained from eSignal, but I can't vouch for the numbers. A slight problem is the date format of the tabular data, which has been conveniently scambled, although legible.