I made a little spreadsheet to calculate the risk of an unrecoverable read error (with input given as the manufacturer's specified rate and the size of the drive in megabytes) and it says that the risk of such an error on a drive exactly 4.0TB with a typical rate of 1 URE per 10^14 bits read is a whopping 27%, no joke. 
I don't know how precise the math is, but that's undoubtedly scary. Working down in capacity...
3.0TB: 21%
2.0TB: 15%
1.5TB: 11%
1.0TB: 7.7%
750GB: 5.8%
640GB: 5.0%
500GB: 3.9%
400GB: 3.1%
320GB: 2.5%
250GB: 2.0%
200GB: 1.6%
160GB: 1.3%
120GB: 0.95%
80GB: 0.64%
40GB: 0.32%
I'm surprised the thing about multiple seeks in quick succession wasn't pointed out to PCBONEZ. If it was, it could have been the end to a pair of flame wars here involving SSDs. (Well, at least the performance half...)
I don't know how precise the math is, but that's undoubtedly scary. Working down in capacity...
3.0TB: 21%
2.0TB: 15%
1.5TB: 11%
1.0TB: 7.7%
750GB: 5.8%
640GB: 5.0%
500GB: 3.9%
400GB: 3.1%
320GB: 2.5%
250GB: 2.0%
200GB: 1.6%
160GB: 1.3%
120GB: 0.95%
80GB: 0.64%
40GB: 0.32%
I'm surprised the thing about multiple seeks in quick succession wasn't pointed out to PCBONEZ. If it was, it could have been the end to a pair of flame wars here involving SSDs. (Well, at least the performance half...)