At a conference I was at last week, an expert suggested that you should try at least two, and preferably three, force-fields. If you get the same result, then you have good confidence that everything is working well. If your result depends on the force-field, then you would have to be suspicious.
Since we don't _really_ know what the correct terms for the van der Waals potential are (and this is pretty fundamental to protein folding), all of the force-fields necessarily involve a degree of approximation.