>>If it's a regular, repeating geometric pattern on the scale of your average mouse motions, that's one of the problem surfaces. Also solid colour (especially black) is not good either.
>>
>>Maybe you could get a mousepad with a fake wood grain or autumn leaves pattern - something non-repeating with lots of features for the mouse sensor to track.
>>
>>The mouse pad I use is excellent - a slightly rough plastic surface embedded with thousands of randomly scattered metal flakes.
>>
>>One last thing to check is that there is not a small piece of lint stuck in the laser sensor - that can make the cursor jump around, too.
>
>What makes it better for such a mouse to be on a pattern surface than a solid one?
>
>Basically, it seems to be something in regards to that and not about the wireless.

http://computer.howstuffworks.com/question631.htm

The mouse takes pictures of the surface, and uses DSP to determine how the pictures move with time (and therefore how the mouse has moved). It can do this only by finding "features" in the pictures, and tracking their motion.

If you have a perfectly smooth, uniform-colour surface, there are no features to track and the mouse goes haywire. The only reason your mouse worked at all on the solid black surface you were trying is because of roughness and imperfections at a small or microscopic scale, and/or surface dust.

If you have a geometric pattern that repeats on various scales, there's a chance the mouse could interpret a picture of a second feature as a motion of the first one, causing the cursor to jump. Dithered images and gradient fills tend to have evenly spaced colour dots that can confuse the mouse, too. A few years back there was an epic fail - I believe at the introduction of VFP7 - MS was handing out free VFP7 mouse pads. Because of gradient fills and colour transitions they were terrible with optical mice. I think I posted about it here, and someone from MS (might have been Mike Stewart) replied saying it was a "known issue" :)

Some Dell mousepads from a few years back had an etched honeycomb/herringbone surface, which would cause optical mice to jump at cell boundaries. I think they were purposely designed to provide good traction for the rubberized balls of mechanical mice.

A random, distinct pattern is best for optical mice. Wood grain is nearly ideal and looks good, too.