My two cents:<p>If we are talking about 1970s and '80s machines, they tend to run pretty hot, related to the PSU technology. Things like electrolytic capacitors "don't like this" and may contribute to cascading failures. PSU failures are apt to fry some chips, some of which may be hard to come by. So, better keep them cool = off.<p>If you consider machines from the 2000s retro, continuous run, avoiding the stresses caused by system start, may help keeping them alive. (I've a MacPro running since 2008, with no failures, apart from failing 3rd party ECC DDR3 RAM. Some of this RAM has even failed twice, the RAM originally shipped by Apple is still fine, though. This machine has only been ever off, when I've been on vacation.) – Machines from the 1990s and early 2000s are pretty much the same, but typically suffer from poor capacitors and/or batteries. So…?