There are reliability estimates based on parts. The one I am slightly familiar is a military standard MIL-HDBK-217F https://www.google.com/search?q=+MIL-HDBK-21

There are reliability estimates based on parts. The one I am slightly familiar is a military standard MIL-HDBK-217F https://www.google.com/search?q=+MIL-HDBK-217F&ie=utf-8&oe=utf-8 https://www.google.com/search?q=+MIL-HDBK-217F&ie=utf-8&oe=u... Doing some searching turns up an online calculator: http://aldservice.com/en/Reliability-Software/free-mtbf-calculator.html http://aldservice.com/en/Reliability-Software/free-mtbf-calc... FWIW, my expectation is that the above will give you a very precise, but totally inaccurate, number. The problem is: 1. MTBF (and other reliability estimates) are based on a statistical sample of a large number of items. If your number of items is small(ish), your statistics become useless. 2. MTBF is not actually computable until you have a meaningful number of failures. 3. Failure rates are very poorly understood, so projecting failure rates before they occur is difficult. 4. Projecting the failure rate of a very complex item (recursively - not only a RPi, but the processor and memory in the RPi) based on estimated failure rates of each individual part that makes up the item is a combinatorial nightmare. To support my assertions above, see Feynman's Appendix to the Rogers Commission Report on the Space Shuttle Challenger Accident http://www.ralentz.com/old/space/feynman-report.html http://www.ralentz.com/old/space/feynman-report.html NASA spent a large fortune estimating the reliability of the space shuttle and was off by several orders of magnitude. Granted, the space shuttle was more complex than a RPi, but expect your estimates, based on the above calculations, to be equivalently (in)accurate.