Which is more reliable? npd or vgchartz?

[binary solo]

Equally reliable I'd say. Both take some hard data and then estimate total sales. Sample size isn't everything if the sample you take is skewed. The bigger the sample the less likely it is to be skewed, but it's not guranteed that you will have a more accurate estimate.

When absolute data has come out for a country or region VGC is consistently within an acceptable margin for estimated sales. E.g. a recent report on PS3 and 360 LTD sales in France showed that VGC was incredibly close with 360 and well within 10% for PS3 (or was it within 5%?). VGC and NPD appear to have a consistently acceptable differential. However VGC doesn't report USA only sales so it's not possible to directly compare. Obviously when NPD is higher than VGC for a given month there is a bit of a disconnect (e.g. 360 sales for Nov '09), but that can happen and the absolute figures can still be within an acceptable range.

VGC data is also dynamic in that figures get adjusted its figures as more data comes in over time (see WE 20/12 where initial report had PS3>1 million but now it's below 1 million). So unlike NPD VGC data becomes more accurate with time.

Also HW sales reliability > SW sales reliability for both NPD and VGC I think.

For the purposes of sales discussions on this website though VGC data is the official data, unless you want to talk USA only sales in which case you have to use NPD for lack of US only VGC numbers. (I'd like to see Americas HW broken down a bit further, like it is with Others).