The UK Government use the British Energy Torness Environmental Impact Statement and the Vattenfall Environmental Impact Statement for their figure of 6gCO₂/KWh. However the figures given in these statements are very questionable.

Other studies such as the University of Sydney ISA study (http://www.isa.org.usyd.edu.au/publications/documents/ISA_Nuclear_Report.pdf) gives a figure of 65gCO₂/KWh which is a bit different. Savacool et al www.nirs.org/climate/background/sovacool_nuclear_ghg.pdf looking at 103 lifecycle analyses gives a similar figure for the carbon dioxide emissions.
I tend to use the ISA study since it examines various forms of energy production and uses the same methodology for all of them. It is a bit problematic taking CO₂ emissions from wind in one study and comparing it to CO₂ emissions from nuclear in another.
The results of the ISA study are given below:

The ISA study also discusses why their findings are different from other studies including the Vattenfall Environmental Impact Statement

There is another study by Storm van Leeuwen. This is somewhat controversial and even though I think it has been wrongly criticised I would not use it. However, the Oxford Research Group and CND have used this study.

The long time period for planning and construction of a nuclear plant is also an important factor. During this time the nuclear plant is not cutting CO₂ emissions. It has been estimated by Jacobson (http://www.stanford.edu/group/efmh/jacobson/PDF%20files/ReviewSolGW09.pdf) that this more than doubles nuclear’s CO2 footprint (from 7-70gCO₂/KWh to 68–180.1gCO2/KWh).

Here is an exampple of how you would work out the CO₂ emissions due to this delay:

• Let us consider a power plant producing 1 KW of power (see here for more about Watts/ KWatts etc)
• In one year this is 8760 KWhr (1x24x365)o f energy.
• If we assume that during the time that the plant is being constructed that gas is being burnt this leads to CO₂ Emissions of 435 g/KWhr. (Gas is assumed to be 500g/KWhr while nuclear is 65g/KWhr).
• This is an extra 3810600 g CO2 per year (435×8760).
• What we need to do now is to spread this out over the lifetime of the plant which we assume is 40 years.
• This is equivalent to an extra 95265g CO2 per year (3810600/40).
• That is 10.88g/KWhr for every year of construction (95265/8760). If we assume that a nuclear plant takes 6 years to construct then that is an extra 65g/KWhr.

Of course we would have to do the same for wind or PV. However, these tend to have much lower construction times than nuclear. If we say that they take 3years then nuclear takes 3years longer. Therefore we would have to add on 32.5g/KWhr to nuclear when comparing it to wind or photovoltaics.

Wikipedia