I prefer the raw recovery available with Sony and Panasonic sensors. The DxO DR measurements seem pretty accurate as far as I can tell. Is this Top 10 just designed to make Canon fanboys feel better about their bodies?

Agree with the two above. Also, of f***ing coarse the DxO sensor scores ignore video ISO performance and JPEG performance, they have nothing to due with the sensor.

There is a flaw in the way DxO adjusts dynamic range figures to account for a sensor's resolution.

Recall that dynamic range is range of brightness that the camera is capable of sensing, expressed as a ratio of the brightest white before clipping to the dimmest grey that is distinguishable from black. DxO also considers only grey values that exceed a certain noise threshold.

To account for resolution, they scale the dynamic range by a factor of the square root of the ratio of the sensor's resolution to a reference resolution of 8 megapixels. That's the theoretical decrease in noise you get by resampling from the camera's resolution to 8 megapixels. The effect is to inflate the dynamic range numbers of high resolution sensors.

The problem is that the inflation may yield a number that exceeds the actual dynamic range of the sensor. No matter how much you downsample, there is still a lower threshold of sensitivity, due to limitations of the sensor's brightness measurement resolution and the resolution of the brightness scale (raw bit depth). For example, a raw linear 14-bit scale has 16384 levels, or 14 stops. But no matter how much you reduce the noise in the lower levels by downsampling, there's still no discrimination finer than between level 0 (or wherever the black level is) and level 1. Yet we routinely see DxO producing dynamic range figures of greater than 14 stops on cameras with 14-bit linear raw images.

In other words, at a certain point noise is no longer the limiting factor in the dynamic range measurement, yet DxO continues to boost the dynamic range figures with no limit for high resolution cameras. Their methodology made sense years ago when noise was always the limiting factor. But now we have cameras with higher dynamic ranges, better low-light noise performance, and very high resolution. And so the figures they are producing are false in many cases.