Hmm, this experience sounds quite like ours.
Summary: bought a pair of servers about 18 months ago, near-identical setup (subtly different mobo revs, 3ware card h/w and f/w revs), each has a 3ware 9500S-12 with 12 Maxtor Diamondmax 10 300GB SATA units (6B300S0).
Early bad things included loss of single/multiple drives at power-up (and even at warm-boot sometimes I think), as many as 8 of the 12 once(!). Both systems have exhibited pretty similar problems.
Since then we've swapped out the 3ware cards for newer ones, upgraded the firmware, reseated the controller end of the cables (drive end is impossible to get to without dismantling, dodgy design). No recent drive loss events on power-up but we power them up only about every 6 months anyway so lousy stats.
Continuing bad symptoms: random loss of drives. Every month or two, a drive will just drop out of the array. This must have happened about 10 times now, and on ONE occasion only, the drive was actually faulty (the Maxtor PowerMax tool said it was bust). All the other times, the drives have been fine, without even an event in their SMART logs (no errors, reallocations, anything).
Sometimes the controller said it kicked them out because of a time-out, sometimes because of an ECC error.
Just asked Maxtor today (their EMEA section) after reading this thread, they said "no known problems except with NForce4".
I had been suspecting bad cabling or PSU issues, but now I'm not so sure it's not the drives failing to play nicely with the controller. (We previously asked 3ware, but they didn't recognise the symptoms (!) and the drives we're using are on the HCL anyway.)
Not sure what to do next... Suggestions welcomed :-)