What to make of sensor rankings?

[Phrasikleia]

Perhaps the DxOMark Sensor rankings (http://www.dxomark.com/index.php/eng/DxOMark-Sensor) site is old hat to many of you, but I recently stumbled upon it and was surprised by some of the rankings.

I was not surprised to see the Nikon D3 take first prize overall for sensor quality, but other scores are not what I expected. For example, the Nikon D300 did not score so well in the "Low-Light ISO" category. It came in #21, scoring worse than the Canon 1000D/XS (which came in at #18). I thought low-light ISO performance was the D300's strong point. The D700 took first place in that category.

I was also surprised to see the Samsung GX 20 get one notch higher than the Pentax K20D, since they use the same sensor. Strange.

Also, in the category of "Dynamic Range", the first three slots are taken by cropped-sensor cameras. Not what I would expect.

The site claims these are objective scores, based on measurable results. I'm curious to know how reliable the scores might be. Any ideas or opinions?

[cube]

The first two slots in the Dynamic Range table are occuppied by SuperCCD SR cameras. This is to be expected.

There are some surprises in the other results, but given that this was measured by DxO, I believe them.

[PCMacUser]

I wouldn't be too bothered by the sensor rankings.

I use a EOS 40D these days, after using a Digital Rebel for a 3-4 years. According to that website, the 40D has a far, far better sensor than the Rebel. This may be true, but it's not very apparent in the photos.