17 biopsies of denervated facial muscles, the zygomatic, the orbicularis oris and the levator labii muscles, showed atrophic myofibers in most cases. There was loss of fiber typability when applying the NADH, the MAG and the alkaline ATPase reactions. The acid ATPase preparations allowed differentiation of myofibers into type I and type II without subtypes. Contrary to normal facial muscles that are richly endowed with motor endplates, no neuromuscular junctions were observed in denervated muscle fibers except one example which might have been obtained by false sampling from the marginal area of denervation or might be the result of partial reinnervation due to sprouting axons from the neighborhood. There was no correlation between the degree of muscle fiber atrophy and the duration of the paralysis. However, fibrosis corresponded to length of denervation. The presence of highly atrophic muscle fibers even 36 years after denervation indicates that the final aim of facial nerve surgery, namely the reinnervation of denervated facial musculature may still be achievable. However, endomysial and perimysial fibrosis may have a considerable impact on the final outcome of such facial nerve surgery. Unsatisfactory correlation between morphological and clinical as well as electromyographical findings in denervated facial muscles requires individual morphological study of each biopsy to assess the probable outcome of reconstructive facial nerve surgery. It therefore appears reasonable even in long-standing facial paralysis, to biopsy denervated facial muscles before or during surgical reanastomosation of the facial nerve. This study provides hints that morphological examination of denervated facial muscles may supplement clinical, electrophysiological, and possibly biochemical diagnostic findings.