Andy Vecsey is right. It is actually better to torque the wheels off the ground. This way there is no other force acting on it.
The manual recommends 80lb-ft. torque, but how tight is right?
During tightening, any threaded fasteners is subjected to 2 different stresses. First, the tension stress set up by actual stretching of the bolt as is tightened. This stress does not develop until the bearing face of the bolt/nut has contacted the work face. Second is the shear stress that is caused by friction between the male and female thread and between the undersurface of the bolt head and the work face.
The tension stress is what we are looking for. This stress will remain in the bolt after the the wrenches are removed, and as long as the total level of stress is within the elastic limits of the bolts. This stress will exert a strong clamping force on the assembly forever, assuming the assembly is rigid.
The shear stress is undesirable but unavoidable. It is undesirable because it is unpredictable. Fortunately this stress goes away almost as soon as we take the wrench off, well for most cases. The amount of friction involved, and therefore the torque required to produce a given amount of tension, varies with the cleanliness of the threads, the fit, the type, the plating, and the amount/type of lubricant used. This makes it difficult to determine the amount of torque required to produce a given amount of stress and resulting strain within the bolts.
This means that most fasteners are not tightened enough due to the friction variation between the female and male threads and bearing surface and work face.
Strange as it may seem, it is actually better to overtighten a bolt than to undertighten it.
The recommended tightening torque value of 12mm stud/bolt (which NSX uses) and grade 12.9 using engine oil lube is 79 lb-ft. That being said and because I don't use any lubrication, I like to use the value of 88 lb-ft.
The right way to tighten the nut is also by hand tight all of them, then use the wrench to tighten them slowly, in steps rather than just tightening them to its final torque all at once. By going a little at a time, we are giving the shear stress a chance to relax and we get more accurate reading.
[This message has been edited by Andrie Hartanto (edited 24 February 2002).]