I would like to provide an automotive perspective for this thread. Automotive companies routinely use 3 procedures for setting torque specifications. These include:
1) use a standard value from a table for noncritical applications (example: M6 screw to attach wiring clips to vehicle structure)
2) use calculation for a non-safety-critical, but structural application (example: M10 screw to attach powertrain mounts to vehicle structure)
3) use testing for safety-critical applications (example: M12 nut to attach steering arm to upright)
Automotive companies have spent considerable time developing materials and coatings specifications for fasteners. For example, friction coefficient is tightly controlled, so using calculations like those from VDI 2230 Systematic Calculation of Bolted Joints (see the thread posted by TVP above) can be used with confidence.
I concur with previous comments that tables and calculations cannot be used for critical attachments if there is no knowledge about the important joint variables. However, calculations are very accurate is the variables are known. These include:
joint geometry
joint material properties
fastener geometry
fastener material properties
friction coefficients
input forces
input torque variation