"How long...HOW LONG?!?!?"

A common question faced by those who perform job analyses or develop employee selection processes for a living is, "How often do we have to go through all this?" We're glad you asked.  Believe it or not, Industrial/Organizational Psychologists have asked this question.  Even more surprising, the answers aren't nearly as self-serving as one might expect.  The general consensus indicates that organizations should seek new selection process validation evidence within about five years of the original study.  A paper authored by Schmit, Lundquist and Beckham concludes the following:

"A survey of members of the Society for Industrial and Organizational Psychology (SIOP) was conducted. Experts in test validation (n = 249) estimated the “shelflife” of a validation study is about 5 years. The experts estimated the shelflife of a job analysis study is 5 to 6 years. The shelflife for a cut-score or passing-score study was estimate at 3 to 3.5 years. The results demonstrated some variability by job family and validation study type in these estimates."

Depending on your situation, you might think that's either a very short time or a long time.  Hence, Renegade Psychology feels compelled to offer its own expert opinion on the matter.

If your organization mostly uses structured selection procedures to select employees for positions requiring the exact same skills and requiring use of the exact same tools utilized a decade ago, you probably can get away with waiting longer than 5 years in between job analyses.  For instance, if you are hiring electricians, and the ones you hire today do the same job with the same tools and require the same knowledge required by the electricians you hired 10 years ago, perhaps you can wait longer than 5 years in between job analyses.

However, it's entirely possible that your initial job analysis sucked.  You might have tried to do it yourself, or you might have hired a mediocre vendor the last time around.  Under such circumstances, you might want to consider the idea that your job analysis might require an update sooner than you had hoped it would.

If your organization is small, or is operating with a restricted amount of financial resources (and who isn't these days?), you will have to utilize a high level of judgment and discretion in making these decisions.  But as a general rule, TheAppliedPsychologist feels comfortable saying that you probably want to take at least a quick look-see at least every 5 years or so (depending on how rapidly the job conditions, work environment, and skill-set changes over time), just to make sure your selection process is working as intended.