The HD I mentioned above was still under 1 year warranty (and was replaced by the mfr, no argument).

A second similar one (a little smaller) went 15 months - on a 1 year warranty. It was easy to diagnose, since it just didn't spin anymore.

HD makers quote MTBF (mean time before failure) in the range of about 10,000 to 20,000 hours for cheaper (standard commercial) drives, up to 80,000 to 120,000 for "high reliability" ones. The "means" don't mean a lot though in any individual case.

The main difference between the high reliability ones and the cheap ones seems to be that they run the expensive ones for a few hours to eliminate the ones that will die early. It raises the "mean," since there are fewer early failures to pull the averages down; but it doesn't really affect the failure distribution expected for individual drives - for the ones that make it past the "early failure" period. HD failures seem to be a clump in "first year," a cluster at "2-3 years," and a fairly large bunch of "never." (The "never" sort of depends on the computer not lasting more than 10 or 12 years though.)

For estimating purposes, "24-7" operation for a year is about 9,000 hours, and a 20,000 hour MTBF means (sort of) that a "more than half" of users should have no HD failures in a year of 24-7 use. 8 hour shifts 5 days a week, makes a standard man-year about 2,088 hours (most US company schedules). You can manipulate the statistics to suggest that more than half of users will get at least 5 years without a failure in single shift "workday" use.

None of this means much if you're in the other half of the users.

John